S.Hrg.119-441

1[Senate Hearing 119-441]2[From the U.S. Government Publishing Office]34                                                        S. Hrg. 119-44156                      SHUT YOUR APP: HOW UNCLE SAM7                    JAWBONED BIG TECH INTO SILENCING8                          AMERICANS, PART II910=======================================================================1112                                HEARING1314                               before the1516                         COMMITTEE ON COMMERCE,17                      SCIENCE, AND TRANSPORTATION18                          UNITED STATES SENATE1920                    ONE HUNDRED NINETEENTH CONGRESS2122                             FIRST SESSION2324                               __________2526                            OCTOBER 29, 20252728                               __________2930    Printed for the use of the Committee on Commerce, Science, and Transportation3132                   [GRAPHIC NOT AVAILABLE IN TIFF FORMAT]3334                Available online: http://www.govinfo.gov3536                               ______3738                 U.S. GOVERNMENT PUBLISHING OFFICE394063-989 PDF                WASHINGTON : 20264142       SENATE COMMITTEE ON COMMERCE, SCIENCE, AND TRANSPORTATION4344                    ONE HUNDRED NINETEENTH CONGRESS4546                             FIRST SESSION4748                       TED CRUZ, Texas, Chairman4950JOHN THUNE, South Dakota             MARIA CANTWELL, Washington,51ROGER WICKER, Mississippi                Ranking52DEB FISCHER, Nebraska                AMY KLOBUCHAR, Minnesota53JERRY MORAN, Kansas                  BRIAN SCHATZ, Hawaii54DAN SULLIVAN, Alaska                 EDWARD MARKEY, Massachusetts55MARSHA BLACKBURN, Tennessee          GARY PETERS, Michigan56TODD YOUNG, Indiana                  TAMMY BALDWIN, Wisconsin57TED BUDD, North Carolina             TAMMY DUCKWORTH, Illinois58ERIC SCHMITT, Missouri               JACKY ROSEN, Nevada59JOHN CURTIS, Utah                    BEN RAY LUJAN, New Mexico60BERNIE MORENO, Ohio                  JOHN HICKENLOOPER, Colorado61TIM SHEEHY, Montana                  JOHN FETTERMAN, Pennsylvania62SHELLEY MOORE CAPITO, West Virginia  ANDY KIM, New Jersey63CYNTHIA LUMMIS, Wyoming              LISA BLUNT ROCHESTER, Delaware6465                 Brad Grantz, Republican Staff Director66           Nicole Christus, Republican Deputy Staff Director67                   Lila Harper Helms, Staff Director68                 Melissa Porter, Deputy Staff Director6970                            C O N T E N T S7172                              ----------73                                                                   Page74Hearing held on October 29, 2025.................................     175Statement of Senator Cruz........................................     176Statement of Senator Cantwell....................................     377Statement of Senator Fischer.....................................    3578Statement of Senator Klobuchar...................................    3779Statement of Senator Blackburn...................................    3980Statement of Senator Lujan.......................................    4181Statement of Senator Schmitt.....................................    4482Statement of Senator Rosen.......................................    4683Statement of Senator Moreno......................................    4884Statement of Senator Curtis......................................    5085Statement of Senator Markey......................................    538687                               Witnesses8889Markham Erickson, Vice President, Government Affairs and Public90  Policy, Google.................................................     591    Prepared statement...........................................     692Neil Potts, Vice President, Public Policy, Meta Platforms Inc....    1093    Prepared statement...........................................    1294Will Creeley, Legal Director, Foundation for Individual Rights95  and Expression.................................................    1496    Prepared statement...........................................    1697Harold Feld, Senior Vice President, Public Knowledge.............    1798    Prepared statement...........................................    1999100                                Appendix101102Response to written questions submitted to Markham Erickson by:103    Hon. Ted Cruz................................................    59104    Hon. Maria Cantwell..........................................    63105Response to written questions submitted to Neil Potts by:106    Hon. Ted Cruz................................................    65107    Hon. Maria Cantwell..........................................    68108Response to written questions submitted to Will Creeley by:109    Hon. Ted Cruz................................................    69110    Hon. Maria Cantwell..........................................    72111Response to written questions submitted to Harold Feld by:112    Hon. Maria Cantwell..........................................    76113114                      SHUT YOUR APP: HOW UNCLE SAM115                    JAWBONED BIG TECH INTO SILENCING116                          AMERICANS, PART II117118                              ----------119120                      WEDNESDAY, OCTOBER 29, 2025121122                                       U.S. Senate,123        Committee on Commerce, Science, and Transportation,124                                                    Washington, DC.125    The Committee met, pursuant to notice, at 10:36 a.m., in126room SR-253, Russell Senate Office Building, Hon. Ted Cruz,127Chairman of the Committee, presiding.128    Present: Senators Cruz [presiding], Fischer, Blackburn,129Schmitt, Curtis, Moreno, Sheehy, Cantwell, Klobuchar, Markey,130Rosen, Lujan, and Fetterman.131132              OPENING STATEMENT OF HON. TED CRUZ,133                    U.S. SENATOR FROM TEXAS134135    The Chairman. Good morning. The Senate Committee on136Commerce, Science, and Transportation will come to order. In1371984, George Orwell chillingly explored how the control of138speech is the control of the mind.139    In describing this world, Orwell powerfully demonstrated140how an authoritarian government can suppress the ability of a141person to speak, dissent, and think in order to control its142people. Long before Orwell, our founding fathers recognized143this threat as well.144    After witnessing a long train of abuses and usurpations by145the British crown, including the suppression of speech and146thought, 56 men signed the Declaration of Independence,147speaking out in defense of man's natural rights and beginning148the formation of a new free nation built upon one's ability to149speak and dissent against the Government.150    The American founders knew all too well that being unable151to speak out against Government encroachment of your rights152ensures the loss of those rights. The right to speak out is the153foundation of a free society. It gives everyone, whether a154Senator or a citizen, a contrarian or a comedian, the power to155persuade and to hold Government accountable.156    And that is why all of us need to defend and protect the157most powerful weapon we have against the Government's ability158to censor its own citizens, the First Amendment. Censorship159around the world is growing. Even allies like the United160Kingdom are making arrests for so called ``offensive speech'',161like criticism of open borders and peaceful prayer near162abortion clinics.163    President Reagan was right when he declared that freedom is164never more than one generation away from extinction. Here in165the United States, we must uphold free speech and zealously166guard against censorship. Earlier this month, I convened a167hearing to examine how our own Government, particularly the168Cybersecurity and Infrastructure Security Agency, censored in169secret, often through third parties.170    CISA and White House officials jaw-boned Big Tech into171deleting tweets, downranking posts, and deplatforming172conservatives under the guise of ``safety and national173security.'' We heard the stories of Alex Berenson and Sean174Davis, who were personally targeted by the Biden175Administration's apparatchiks for opinions and facts they176shared on COVID and election fraud.177    They not only suffered significant financial loss, but more178importantly, they were silenced by their own Government. Today,179we pick off where the story left off, with witnesses from two180Big Tech companies that were pressured by the Biden181Administration to censor the American people, Google and Meta.182    Today, these same companies claim to zealously advocate for183free speech. While I welcome their newfound religious184conversion, they owe apologies to Alex Berenson, to Sean Davis,185and to countless Americans they silenced at the Biden186Administration's behest. No major tech company, let alone the187ones before us, publicly spoke out when this was happening.188They were silent.189    Despite armies of lobbyists and legal resources without190limit, Big Tech was speechless when the Biden White House191bullied them into censoring regular citizens. Neither Big Tech,192nor any of my Democrat colleagues, spoke up. They seemed quite193content to follow the example of the United Kingdom,194criminalizing speech that some find offensive under the guise195of ``misinformation.''196    But the fact is the Government cannot or at least should197not be the arbiter of truth. You don't get to decide free198speech protections are available only to favored or popular199speech. It is not up to bureaucrats and politicians to decide200which speech is reasonable and which is hateful, and then only201accept that speech that they find to be socially acceptable and202inoffensive. You don't get to cancel someone for using the203wrong pronouns or challenging conventional wisdom on vaccines.204    Now, our commitment to free speech under the rule of law205means protecting speech that some people find unreasonable, or206offensive, or even hateful. So, what does that mean in207practice? Alex Berenson and Sean Davis can prove that the Biden208Administration was behind their online censorship, yes, but209they still face significant legal obstacles in recouping their210losses.211    Their stories and countless others like them have revealed212dangerous gaps in our legal protections for free speech. It213cannot be that their First Amendment rights go un-remedied214merely because the violators are no longer in office or because215Congress has withheld a damages remedy.216    This collusion between big Government and Big Tech is a217warning that we need stronger tools to guard our cherished218freedoms. That is why I will soon introduce new legislation,219the Jawbone Act, which will provide a robust right to redress220when Americans are targeted by their own Government.221    I am hopeful that today's hearing will reflect bipartisan222interest in protecting free speech and finding common ground on223measures like this one. We cannot remain idle in the face of224Government agencies manipulating private companies into225silencing their critics.226    We must act decisively to uphold the First Amendment, to227protect open debate, and to defend every American's right to228speak freely without fear of Government reprisal. I will now229turn to Ranking Member Cantwell.230231               STATEMENT OF HON. MARIA CANTWELL,232                  U.S. SENATOR FROM WASHINGTON233234    Senator Cantwell. Thank you, Mr. Chairman, and thank you235for having this hearing, for the witnesses being here. But my236fundamental question still is, where is Brendan Carr? We might237have the right hearing, but I am not sure we have the right238witnesses.239    We might have the right questions, but I am not sure we240have the right Administration we are calling into question. I241will take the opportunity to ask the witnesses about AI, about242algorithms, about competition on platforms.243    But as we discuss censorship today, this hearing isn't244serious if it ignores the ongoing corporate consolidation green245lighted by the Trump Administration, which is only approving246deals that come with a political pro quo. Mr. Feld, in your247testimony, you talk about, ``consolidation amplifies the248ability to control content on an unprecedented scale.''249    And I appreciate that you recognize the threat posed by250corporate consolidation to free speech and independent251journalism. In fact, you also say, ``a handful of opaque252algorithms are controlling the majority of discussion online.''253    Just this year, FCC Chairman Carr approved Skydance's $8254billion acquisition of Paramount, but only after Paramount paid255$16 million to settle with President Trump and install a first-256term Trump appointee as the ombudsman at CBS.257    And FTC Chairman Ferguson approved Omnicron's $13 billion258acquisition of Interpublic, creating the world's largest media259buying agency while ignoring the serious concerns about market260consolidation.261    So, rather than protecting consumers and competition, the262FTC imposed conditions to favor certain kinds of content. The263FTC later in complaints--as Mr. Feld is saying in his264testimony, it is really important to have somebody at the FTC265that is making sure that we understand the consumer harm when266you reduce competition, when you basically are reducing267investment.268    The consolidation of audiences and advertising dollars on269just a few social media platforms pose real risk to democracy,270to the sustainability of local, diverse, and trusted media271sources.272    When a handful of companies control how information273spreads, that makes it easier for any Administration to abuse274that power with political threats to chill speech. What we have275seen, both Meta and Google recently reached a $25 million276settlement with President Trump.277    Those settlements raise concerns about the kind of back278channel arrangements that can distort information. The stakes279are high now because as we look to turning to AI for the280future, how will that tool be used? A July poll from the281Associated Press show that 60 percent of Americans use AI to282find information.283    While AI is a tremendous power for good, this trend raises284concerns of the concentration and monopolization of that285information. The amount of control in the hands of just a286select few could erode the foundation of free speech.287    The First Amendment isn't just a line in the Bill of288Rights. It is about an operating system for our democracy that289protects everybody, comedians, protesters, the free press, and290obviously people we disagree with. That is why this committee291should focus on ensuring competition, diverse and truly292independent media systems, and all supportive of both293transparency expressed in this panel's testimony today and in294the previous panel's testimonies.295    Transparency in process can be a very good antiseptic. The296proposed TikTok deal is a perfect example of the danger when297those principles break down. It places yet another major media298platform under control--already Paramount, CBS, Skydance, and299reportedly those seeking to buy Warner Brothers' Discovery.300What are the details of how we are going to rid the Chinese301influence of these algorithms and their misuse of algorithms? I302am still waiting for that explanation.303    So, we have seen how consolidation erodes both competition,304and we only have a handful of companies owning outlets that305Americans rely on for news, entertainment, and those companies306become easier to intimidate.307    So I ask again, where is Chairman Carr? These are important308issues. I hope they will appear before this committee at some309point in time, because we need to understand the direction that310we are going in our country to protect free speech.311    And that is why the work of this committee should be, as312you said, Mr. Chairman, to defend the First Amendment, and have313a free, diverse media landscape. It is more urgent than ever.314Thank you.315    The Chairman. Thank you very much. And I would note the316Ranking Member asked where Brendan Carr is. As she knows, this317committee has already announced we will be having oversight318hearings, and Brendan Carr will be testifying before this319committee.320    And indeed, in the preceding two years, when the Democrats321had the majority on this committee, we had precisely zero322oversight hearings of the FCC. We had zero oversight hearings323of the FTC. And I haven't announced it, but we will be having324oversight hearings of the FTC as well, because this committee325has a responsibility to engage in oversight.326    So Mr. Carr will be here, and each member of this committee327on both sides of the aisle will be able to question him on any328policies or questions they like. I now want to introduce our329witnesses for today. Our first witness is Markham Erickson, the330Vice President of Governmental Affairs and Public Policy at331Google.332    Mr. Erickson leads Google's Government Affairs and Public333Policy Centers of Excellence, a global team focused on applying334law to technology and the internet. Our second witness is Neil335Potts, the Vice President of Public Policy at Meta. Mr. Potts336has worked for Facebook, or Meta, since 2016, and I look337forward to hearing his testimony about Meta's experience with338jawboning.339    Our third witness is Will Creeley, Legal Director for the340Foundation for Individual Rights and Expression. His experience341advocating for free expression spans five Presidential342Administrations, and his writings have been featured in343prominent newspapers and Supreme Court briefs.344    Our final witness is Harold Feld, Senior Vice President of345Public Knowledge. Mr. Feld has practiced law for more than 20346years at the intersection of technology, broadband, and media347policy. Mr. Erickson, you are recognized for five minutes.348349   STATEMENT OF MARKHAM ERICKSON, VICE PRESIDENT, GOVERNMENT350               AFFAIRS AND PUBLIC POLICY, GOOGLE351352    Mr. Erickson. Chairman Cruz, Ranking Member Cantwell, and353members of the Committee, thank you for the opportunity to354speak with you today. My name is Markham Erickson, and I serve355as Vice President of Government Affairs and Public Policy at356Google.357    Google's mission, from its inception to today, has been to358organize the world's information and make it universally359accessible and useful. Founded in America, we proudly advance360this mission around the globe.361    And while our platforms continue to experience tremendous362success worldwide, it is hard to imagine that our story could363have begun anywhere else but here in America, where the364commitment to freedom of expression is so profound. And yet,365even in America and certainly in many other countries around366the world, we often find this commitment challenged by others.367    We hear from Government agencies, legislators, NGOs,368academics, and individual users who inform us of content that369they believe we should remove, or on the other hand, content370that we have removed that they believe should remain available.371This happens across the political spectrum and around the372world, and it has for years.373    It is not new to us. With respect to the governments in374countries where we operate, we have a responsibility to engage375on this issue, just as I am doing here today. While we are a376company dedicated to the goal of making the world's information377universally accessible, that doesn't mean that we don't have378certain rules.379    Terrorist content, child sexual abuse material, hate380speech, and other harmful content have no place on our381platforms. To be crystal clear, we develop and enforce our382policies independently. That is a right we enjoy in this383country as a matter of law, and one we maintain with purpose384and determination.385    We recognize our responsibility, and we have policies386against content that we conclude is illegal, offensive, or387dangerous. We independently and voluntarily evaluate whether388content violates our terms of policy--our terms of services and389policies, reach our own conclusions and take enforcement action390in accordance with our own determinations.391    And sometimes that means we tell the Government no. During392the pandemic, officials in the Biden Administration pressed393Google to remove certain COVID-19 content on YouTube. Despite394this outreach, we continued to develop and enforce our policies395independently, rejecting suggestions that did not align with396those policies.397    And today, we exercise this same independence over our398policies and products under the current Administration and with399governments around the world. Outside of the United States, we400have fought that fight at breathtaking cost.401    Multiple governments have shut down access to our services402when we would not remove content that complied with our403policies. As one example, when Russian authorities ordered404YouTube to remove political videos critical of the regime, we405refused. The government threatened our employees with jail and406fined us.407    Today, those fines exceed more than the world's GDP, yet we408continue to resist. We also recognize that transparency409regarding Government actions--interactions with companies is410essential to foster public trust and uphold the principles of411freedom of expression. Google publishes data in our biannual412transparency report on Government content removal requests.413    We welcome the Committee's consideration of legislation414that would create more transparency in how the U.S. Government415engages with industry, and Chairman Cruz, we support in416principle some of the legislative recommendations from your417staff in your report. As we make our independent choices about418how we moderate content, we recognize that speech is more419important than ever.420    We have to get this right. Think of the sheer scale. Google421responds to billions of search queries every day and over 500422hours of content are uploaded on YouTube every single minute,423reflecting the vast proliferation of viewpoints and a range of424perspectives. This scale translates to thriving economic growth425in the United States.426    As website publishers and YouTube creators find success,427they build businesses, hire employees, and contribute to their428communities while creating content for billions of people429around the world. The opportunity to help drive a thriving430digital ecosystem is only just beginning.431    The scientific innovations made possible by artificial432intelligence are creating a generational technology shift.433Google's AI advances in predicting protein structures, for434example, have saved at least 400 million years of research and435are helping 3 million scientists in 190 countries make progress436in addressing things like cancer and malaria. Google uses AI to437forecast floods, detect wildfires, and provide tools to improve438farming efficiency and boost crop yields.439    These discoveries, this scientific progress is made440possible by America's tradition of protecting the free and441robust expression of ideas. We remain committed to doing our442part to keep the digital ecosystem safe, reliable, and open to443free expression. We rely on the trust of our users, and we444intend to keep on earning it. I look forward to your questions.445    [The prepared statement of Mr. Erickson follows:]446447 Prepared Statement of Markham Erickson, Vice President of Government448                   Affairs and Public Policy, Google449    Chairman Cruz, Ranking Member Cantwell, and Members of the450Committee, thank you for the opportunity to speak with you. My name is451Markham Erickson, and I serve as Vice President of Government Affairs452and Public Policy at Google.453    Google's mission--from its inception to today--has been to organize454the world's information and make it universally accessible and useful.455Founded in America, Google proudly advances this mission around the456globe. And while our platforms continue to experience tremendous457success worldwide, it is hard to imagine that our story could have458begun anywhere else but here in the United States, where the commitment459to freedom of expression is so profound.460Independent Policy Development and Enforcement461    We take seriously the importance of protecting free expression and462access to a range of viewpoints, while also maintaining and enforcing463responsible policy frameworks. Terrorist content, child sexual abuse464material, hate speech, and other harmful content have no place on our465platforms. When it comes to the manner in which we moderate the content466on our platforms, we are crystal clear: we develop and enforce our467policies independently. We independently and voluntarily evaluate468whether content violates our terms and policies, reach our own469conclusions, and take enforcement action in accordance with these470determinations.471    Around the world, we hear from government agencies, legislators,472non-governmental organizations, academics, and individual users who473inform us of content that they believe we should remove, or, on the474other hand, content that we have removed that they believe should475remain available. This happens across the political spectrum and around476the world, and it has for years. We have a responsibility to engage on477these issues in the countries where we operate.478    Sometimes, the results of our independent evaluations mean that we479tell a government ``no'' and we decline to remove content as they480request. For example, during the pandemic, officials in the Biden481Administration pressed YouTube to remove certain COVID-19 content.482Despite this outreach, we continued to develop and enforce our policies483independently, rejecting suggestions that did not align with our484policies. Today, we exercise this same independence over our policies485and products under the current Administration and with governments486around the world.487Google's Interactions with Governments Around the World488    Outside of the United States, we also have pushed back against489requests to moderate content, even at great cost. Multiple governments490have attempted to shut down user access to our services when we would491not remove content that complied with our policies.492    As one example, when Russian authorities ordered YouTube to remove493political videos critical of the regime, we refused. The government494threatened our employees and fined us. Kremlin-aligned media outlets495sued us in Russia, demanding we provide a platform for them. After we496refused, the Russian government imposed fines against us that forced us497to leave the market. Today those fines exceed more than the world's498GDP. We continued to fight back, bringing suit in the European Court of499Human Rights and securing a ruling that Russia had violated freedom of500expression. We have since filed successful court cases in the United501Kingdom and the United States to stop the media outlets from enforcing502these unjust fines against us outside Russia.503    While Google has a proud track record of pushing back against or504objecting to overly broad or otherwise inappropriate government demands505for content removals, we recognize that communication with law506enforcement and the government is a critical component of our efforts507to keep people safe on our platforms and services. Google works with508government officials, law enforcement entities, and legislators across509the political spectrum in numerous ways to assess threats and to510counter attempts to deceive, harm, or take advantage of users.511    We have processes in place to swiftly refer threats of imminent512harm--including those involving terrorism, mass shootings, violent513events, or other urgent incidents--and specified illegal activity514occurring on our platform to law enforcement. We also have procedures515for our various internal product teams, as well as external sources, to516escalate potential threats and criminal activity they see on our517platforms, including child sexual exploitation. Google personnel assess518escalated threats and refer matters to law enforcement whenever519appropriate, consistent with due process and privacy protections.520    In addition to these urgent referrals, we maintain regular521communication channels with law enforcement, government entities, and522industry partners as part of our efforts to keep people safe and523understand and adapt to trends and new forms of abuse. We rely on524information learned through such channels to ensure the integrity of525our products. When appropriate, we share threat intelligence, provide526criminal referrals, and hold follow-up briefings. More information on527our threat intelligence work can be found here [https://528cloud.google.com/security/products/threat-intelligence?hl=en] and here529[https://blog.google/threat-analysis-group/].530    Courts and government agencies around the world regularly ask531Google to remove information from Google products for alleged violation532of law. We review these legal demands closely to determine if content533should be removed because it violates our terms of service and content534policies, or should be restricted within a specific country because it535violates a local law. In order for us to evaluate a legal demand from a536government entity, it must provide a clear explanation of how the537content violates controlling laws or regulations or our policies.538    We do not always remove content in response to a government's legal539demand. Some legal demands may not be specific enough for us to know540what the government wanted us to remove (for example, no URL is listed541in the request) or lack sufficient explanation of why the government542believes content violates a given law.543    We also know that transparency regarding government interactions544with companies is essential to foster public trust and uphold the545principles of free expression. Google publishes data in our biannual546Transparency Report on government content removal requests--part of our547efforts to inform discussions about and shed light on these requests.548    We welcome the Committee's consideration of legislation that would549create more trust and transparency in how the U.S. government engages550with industry, and we support in principle some of the legislative551recommendations outlined in the Committee's recent report.552Opportunity and Innovation553    As we make our independent choices about how we moderate content,554we recognize that speech is more important than ever. Google responds555to billions of search queries every day, and over 500 hours of content556is uploaded on YouTube every single minute, reflecting a vast557proliferation of viewpoints and a range of perspectives. This scale558translates to thriving economic growth in the United States. As website559publishers and YouTube creators find success, they build businesses,560hire employees, and contribute to their communities across the country561while creating content for billions of people.562    We recognize that our products have an important role to play in563facilitating and contributing to a healthy digital ecosystem and making564information available to our billions of users. We find ourselves in an565incredibly critical moment in the history of the Internet. The566scientific innovations made possible by artificial intelligence are567creating a generational technological shift. For instance, Google's AI568advances in predicting protein structures have saved at least 400569million years of research and are helping three million scientists in570190 countries make progress in addressing diseases like malaria and571cancer. Google is also using AI to forecast floods, detect wildfires,572and provide tools to improve farming efficiency and boost crop yields.573Our AI projects are driving positive change in areas including health,574accessibility, science, and economic impact.575    Selected examples of these AI applications to address society's576biggest challenges include:577578   Fighting Cancer. Artificial Intelligence holds the promise579        of revolutionizing humanity's fight against cancer and other580        diseases. At Google, we are particularly proud of our recent581        advancements, including:582583     Childhood leukemias and many other cancers have584            incredibly complex genetic signatures, requiring tailored585            treatments based on their specific mutations. DeepSomatic586            is an AI-powered tool that helps scientists and doctors587            spot genetic variants in cancer cells. Our partners at588            Children's Mercy in Kansas City used DeepSomatic to589            identify 10 new genetic variants in samples of childhood590            leukemia that had been missed by previous techniques. If591            they can pinpoint how and why a particular form of cancer592            is affecting a patient, they may be able to develop593            personalized cures. Remarkably, DeepSomatic can also594            generalize to cancers it hasn't seen before. For example,595            without any training on the brain cancer glioblastoma,596            DeepSomatic was able to pinpoint which genetic variants597            cause it. This suggests it could work even on rare or new598            types of cancer--a big milestone that marks 10 years of599            genomics research at Google. We are also working with600            clinicians, patients, and partners to build an AI system601            for mammography, which could help radiologists detect602            breast cancer more accurately, quickly, and consistently.603604     A major challenge in cancer immunotherapy is that many605            tumors are ``cold''--invisible to the body's immune system.606            A key strategy to make them ``hot'' is to force them to607            display immune-triggering signals through a process called608            antigen presentation.'' Recently, in a collaboration with609            Yale University, Google DeepMind developed an AI model that610            was able to generate a new hypothesis about interventions611            that could boost antigen presentation--and laboratory612            testing revealed that the hypothesis was sound, ``resulting613            in a roughly 50 percent increase in antigen presentation,614            which would make the tumor more visible to the immune615            system. With further preclinical and clinical validation,616            such hypotheses may be able to ultimately accelerate the617            path to new therapies.618619   Predicting Proteins Structures with AlphaFold. For decades,620        scientists tried to find a method to reliably determine a621        protein's structure from its sequence of amino acids alone.622        These methods included nuclear magnetic resonance and X-ray623        crystallography, which relied on extensive trial and error,624        years of painstaking work, and multi-million-dollar specialized625        equipment. This grand scientific challenge is known as the626        protein-folding problem. In 2016, researchers at Google627        DeepMind began work on this problem. We built an AI system by628        teaching it through visual sequences and structures of around629        100,000 known proteins. Now, this system can predict the shape630        of all 200 million known proteins, almost instantly, down to631        atomic accuracy. We call this system AlphaFold. Today,632        AlphaFold has revealed millions of intricate 3D protein633        structures, and is helping scientists understand how life's634        molecules interact. For this breakthrough, Co-founder and CEO635        of Google DeepMind and Isomorphic Labs Sir Demis Hassabis, and636        Google DeepMind Director Dr. John Jumper were co-awarded the637        2024 Nobel Prize in Chemistry.638639   Facilitating Human Communication. Google is committed to640        facilitating communication. For example, Google's Project641        Relate is an Android app that helps individuals with non-642        standard speech communicate more easily. By offering features643        like speech-to-text transcription and clear voice repetition,644        it enables users to express themselves and manage daily tasks645        effectively. Google also is committed to developing language646        inclusion technology. There are over 7,000 languages spoken647        around the world. To date, few of these languages have been648        well represented online. To capture the diversity of how we649        communicate globally, Google has built an AI model to help650        people connect and better understand the world around them.651652   Advancing Earth Sciences. Google is using AI to advance653        earth sciences in a variety of ways such as:654655     Flood Forecasting. Floods are the most common type of656            natural disaster and nearly 1.5 billion people, or some 19657            percent of the world population, are directly exposed to658            substantial risks from severe flood events worldwide. Our659            AI-powered flood forecasting platform can predict when and660            where riverine flooding will occur up to seven days in661            advance. This allowed us to provide flood forecasting in 80662            countries in areas where 460 million people live. Where663            possible, we also provide forecasts in Google Search and664            Google Maps and via Android notifications.665666   Agriculture and Farming. Using Google Research's NeuralGCM667        and other advanced AI models, the University of Chicago's668        Human-Centered Weather Forecasts Initiative built a high-669        quality forecast that is more efficient, more flexible, and670        more accurate than traditional forecasts. In 2025, the program671        successfully delivered tailored, advanced forecasts to 38672        million farmers in India--accurately predicting an unusually673        delayed monsoon season and allowing the farmers to proactively674        adjust their planting decisions accordingly.675676   Wildfire tracking. Our alerts, simulation models, and677        wildfire risk forecasting systems are helping provide accurate678        wildfire information to affected communities and fire679        authorities. In an exciting recent development, the first680        satellite for the FireSat constellation officially made contact681        with Earth. This satellite is the first of more than 50 in a682        first-of-its-kind constellation designed to use AI to detect683        and track wildfires as small as a classroom (roughly 5x5684        meters). Currently, many wildfire authorities depend on685        satellite imagery that is low-resolution or only updated a few686        times a day. FireSat will advance the science and practice of687        fighting wildfires by providing high-resolution imagery updated688        globally every 20 minutes, helping emergency responders catch689        wildfires before they become destructive.690691   Calling in Our Corals--SurfPerch. Our AI model can expedite692        the process of finding solutions to preserve and restore coral693        reefs, developed in collaboration with Google researchers and694        enriched by contributions from citizen scientists on the695        Calling in Our Corals.696697   Green Light. Our AI tool measures driving trends at698        intersections and develops recommendations that city traffic699        engineers can use to optimize the timing of traffic light700        changes to reduce stop-and-go traffic.701702   Earth AI. The hardest and most important questions in703        planetary science and crisis response are never about just one704        kind of geospatial information--they are about pulling it all705        together. In response, we are developing Earth AI in order to706        knit all of that information and predictive power together.707        Questions that are currently impossible to answer because708        they're too complex and draw on too many disparate geospatial709        resources will become possible to tackle. And this, in turn,710        will prompt new research--new collection of useful data about711        Earth, new kinds of sensors and new uses of AI to model712        sophisticated interconnected patterns across the planet.713714     Advancing Quantum Computing. At Google Quantum AI, we715            believe quantum computers can help shape a brighter future716            and solve otherwise impossible problems in the fields of717            national security, drug discovery, industrial chemistry,718            energy, and more. We are optimistic that within five years719            we will see real-world applications that are possible only720            on quantum computers. Google Quantum AI announced the721            first-ever demonstration of verifiable quantum advantage, a722            significant step towards real world applications for723            quantum computing. We have run an algorithm, which we call724            Quantum Echoes, on our quantum computer that would take725            13,000 times longer on one of the world's fastest726            supercomputers. This advance may help us better understand727            molecular structures, with potential implications for728            fields like drug discovery, medicine, and battery729            development. The United States leads the development of730            quantum computing, and Google has the most advanced, state-731            of-the-art quantum computing effort anywhere in the world.732            We proudly make our chips in the United States, at a733            dedicated superconducting fabrication facility. All of our734            quantum computing technology is located in our U.S. quantum735            computing laboratories.736                                 # # #737    The future is before us. And it is a future made possible by unique738attributes of the United States of America and the freedoms we hold739dear. These discoveries, this scientific progress, and the resulting740economic growth in the United States is made possible by America's741tradition of protecting the free and robust expression of ideas. We742remain committed to doing our part to keep the digital ecosystem safe,743reliable, and open to free expression; we rely on the trust of our744users and we intend to keep earning it.745746    The Chairman. Thank you. Mr. Potts, you are recognized.747748           STATEMENT OF NEIL POTTS, VICE PRESIDENT,749               PUBLIC POLICY, META PLATFORMS INC.750751    Mr. Potts. Thank you, Chairman. Chairman Cruz, Ranking752Member Cantwell, members of the Committee, thank you for the753opportunity to testify today. My name is Neil Potts, and I am a754Vice President for Public Policy at Meta.755    I have been with the company for nearly 10 years, and in my756current role, I am responsible for oversight of the development757and implementation of our community standards, which provide758the rules for what we allow and do not allow on the platform. I759am a graduate of the United States Naval Academy and the760University of Virginia School of Law.761    Prior to joining Meta, I worked as a lawyer in private762practice and served as a ground intelligence officer in the763United States Marine Corps, where I was deployed in support of764Operation Enduring Freedom and Operation Iraqi Freedom.765    At Meta, our mission is to build the future of human766connection and the technology that makes it possible. Key to767this work is our foundational commitment to free expression.768Every day, millions of Americans use our services to stay in769touch with friends and family, generate creative content, and770learn about what is going on in the world.771    We want to give them a voice and make sure they can engage772with their communities in a safe way. The diversity of773viewpoints, expression, and experience on our platforms774highlights much of what is good about Meta, but it also775presents challenges when deciding what content should and776should not be allowed.777    These are complex issues. To create a place that is safe778for a variety of people, we have to make hard decisions about779what should be permitted. Some people think we take down too780much content, others think we down too little content. And781while we know everyone will not agree with every decision we782make, we remain committed to providing transparency about our783content moderation and our enforcement policies, and we have784worked hard to prioritize speech.785    Our goal is always to get it right, but we recognize that786we are not perfect. When there are concerns about how we787approach enforcement, we appreciate hearing from people,788including you here in Congress, about how can we improve. In789that spirit, I appreciate the opportunity to share Meta's790perspective on communications with Government officials.791    We regularly hear from governments around the world and792others with various concerns around public discourse and public793safety. And our job, no matter where content reports may come794from, is to ensure that we make independent decisions based on795our policies.796    As our CEO has previously acknowledged, in 2021, senior797officials from the Biden Administration, including the White798House, repeatedly pressured our teams for months to censor799certain COVID-19 content, including humor and satire, and they800expressed a lot of frustration when our teams did not agree.801    Ultimately, it was our decision to take down content and we802own those decisions, including COVID-19 related changes we made803to enforcement in wake of that pressure. To be clear, we804believe that Government pressure was wrong and wish we had been805more outspoken about it. We should not compromise our content806standards due to pressure from any Administration in either807direction, and we are ready to push back if something like this808happens again.809    We also work to promote free expression, and to that end we810recently implemented a series of significant changes designed811to help bring us back to our roots and allow for more speech.812Among other things, we ended our third-party fact checking813program and moved to a community notes model.814    We got rid of a number of restrictions on topics that are815subject to frequent political discourse and debate. We changed816how we enforce our policies to reduce mistakes, and we adopted817a more personalized approach to political content. These818changes, which are described in my written testimony, are an819attempt to return to our ideals about free expression.820    That means being vigilant about the impact our policies and821systems have on people's ability to make their voice heard, and822having the humility to change our approach when we know we are823getting something wrong.824    We believe these changes are having a positive impact and825look forward to continuing to give people a voice across our826platforms. Senator, Ranking Member, thank you again. I welcome827your questions.828    [The prepared statement of Mr. Potts follows:]829830   Prepared Statement of Neil Potts, Vice President, Public Policy,831                          Meta Platforms Inc.832    Chairman Cruz, Ranking Member Cantwell, Members of the Committee:833Thank you for the opportunity to testify today. My name is Neil Potts,834and I am a Vice President for Public Policy at Meta. I have been with835the company for nearly 10 years, and, in my current role, I'm836responsible for oversight of the development and implementation of our837Community Standards, which provide the rules for what types of content838are and are not allowed on our platforms. I am a graduate of the United839States Naval Academy and the University of Virginia School of Law.840Prior to joining Meta, I worked as a lawyer in private practice and841served as a ground intelligence officer in the United States Marine842Corps and was deployed in support of Operation Enduring Freedom and843Operation Iraqi Freedom.844    At Meta, our mission is to build the future of human connection and845the technology that makes it possible. Key to this work is our846foundational commitment to free expression. Every day, millions of847Americans use our services to stay in touch with friends and family,848generate creative content, learn about what's going on in the world,849and more. We want to give them a voice and make sure that they can850engage with their communities in a safe way.851    The diversity of viewpoints, expression, and experiences on our852platforms highlights much of what is best about Meta. But it also853presents challenges when deciding what content should and should not be854allowed. These are complex issues. To create a place that is safe for a855variety of people, we have to make hard decisions about what should be856permitted. Some people think we take down too much content, while857others think we remove too little. While we know not everyone will858agree with every decision that we make, we remain committed to859providing transparency to our content moderation and enforcement860policies, and we have worked to prioritize speech. Our goal is to861always get it right, but we recognize that we are not perfect. And when862there are concerns about how we approach our enforcement, we appreciate863hearing from people, including you here in Congress, about how we can864improve.865    In that spirit, I appreciate the opportunity to share Meta's866perspective on communications with government officials, which is the867focus of the Committee's invitation that I am prepared to address. We868regularly hear from governments around the world and others with869various concerns around public discourse and public safety. Our job, no870matter where content reports may come from, is to ensure that we make871independent decisions based on our policies.872    As our CEO has acknowledged, in 2021, senior officials from the873Biden Administration, including the White House, repeatedly pressured874our teams for months to censor certain COVID-19 content, including875humor and satire, and expressed a lot of frustration with our teams876when we didn't agree. Ultimately, it was our decision whether or not to877take content down, and we own our decisions, including changes we made878to our COVID-19-related enforcement in the wake of this pressure. We879believe the government pressure was wrong and wish we had been more880outspoken about it. We should not compromise our content standards due881to pressure from any Administration in either direction, and we're882ready to push back if something like this happens again.883Our Approach to Content Moderation884    We moderate content according to our published policies, including885our Community Standards. We enforce these policies in an effort to help886keep people on our platforms safe, reduce objectionable content, and887help people participate on our platforms responsibly. We strive to be888clear and transparent about what our standards are, and we seek to889apply them to everyone consistently. When we identify content that890violates our policies, we take action.891    Over the years, this work led us to develop a set of increasingly892complex systems to manage content across our platforms. These efforts893were well-intentioned, but they went too far. They expanded over time894to the point where we were making too many mistakes, frustrating our895users, and too often getting in the way of the free expression we set896out to enable.897Recent Changes to Our Content Moderation Program898    To address this, we recently implemented a series of significant899changes designed to bring us back to our roots and allow for more900speech. As we said at the time of their announcement:901902   Ending Our Third-Party Fact-Checking Program and Moving to a903        Community Notes Model. When we launched our independent fact-904        checking program in 2016, we were very clear that we didn't905        want to be the arbiters of truth. We made what we thought was906        the best and most reasonable choice at the time, which was to907        hand that responsibility over to independent fact-checking908        organizations. The intention of the program was to have these909        independent experts give people more information about the910        things they see online, particularly viral hoaxes, so they were911        able to judge for themselves what they saw and read.912913    That's not the way things played out, especially in the United914        States. Experts, like everyone else, have their own biases and915        perspectives. This showed up in the choices some made about916        what to fact check and how. Over time we ended up with too much917        content being fact checked that people would understand to be918        legitimate political speech and debate. Our system then919        attached real consequences in the form of intrusive labels and920        reduced distribution. A program intended to inform too often921        became a tool to censor.922923    So we changed our approach and ended the third-party fact checking924        program in the United States, moving to a Community Notes925        program. We've seen this approach work on X--where they empower926        their community to decide when posts are potentially misleading927        and need more context, and people across a diverse range of928        perspectives decide what sort of context is helpful for other929        users to see. We think this is a better way of achieving our930        original intention of providing people with information about931        what they're seeing--and one that's less prone to bias.932933    Meta does not write Community Notes or decide which ones show up.934        They are written and rated by contributing users. Community935        Notes also require agreement between people with a range of936        perspectives to help prevent biased ratings. We intend to be937        transparent about how different viewpoints inform the Notes938        displayed in our apps and are working on the right way to share939        this information.940941    We have phased in Community Notes in the U.S. and continue to942        improve them. We have stopped demoting fact-checked content,943        and, instead of overlaying full screen interstitial warnings944        people have to click through before they can even see a post,945        we now use a much less obtrusive label indicating that there is946        additional information for those who want to see it.947948   Allowing More Speech. We found that we were over-enforcing949        our rules, limiting legitimate political debate, censoring too950        much trivial content, and subjecting too many people to951        frustrating enforcement actions. We want to undo the mission952        creep that has made our rules too restrictive and too prone to953        over-enforcement. We got rid of a number of restrictions on954        topics like immigration and gender that are the subject of955        frequent political discourse and debate.956957    We also changed how we enforce our policies to reduce the kind of958        mistakes that account for the vast majority of the censorship959        on our platforms. We historically used automated systems to960        scan for all policy violations, but this resulted in too many961        mistakes and too much content being censored that should not962        have been. So, we have continued focusing these systems on963        tackling illegal and high-severity violations, like terrorism,964        child sexual exploitation, drugs, fraud, and scams. For less965        severe policy violations, we now rely on someone reporting an966        issue before we take any action.967968    We also found we were demoting too much content that our systems969        predicted might violate our standards. We got rid of most of970        these demotions and now require greater confidence that the971        content violates for the rest. Additionally, we tuned our972        systems to require a much higher degree of confidence before a973        piece of content is taken down.974975    People are often given the chance to appeal our enforcement976        decisions and ask us to take another look, but the process can977        be frustratingly slow and doesn't always get to the right978        outcome. We added extra staff to this work and, in more cases,979        we are also now requiring multiple reviewers to reach a980        determination in order to take something down. We are working981        on ways to make recovering accounts more straightforward and982        testing facial recognition technology, and we've started using983        AI large language models to provide a second opinion on some984        content before we take enforcement actions.985986   A Personalized Approach to Political Content. Since 2021, we987        have made changes to reduce the amount of civic content people988        see--posts about elections, politics, or social issues--based989        on the feedback our users gave us that they wanted to see less990        of this content. But this was a pretty blunt approach. We991        started phasing this back into Facebook, Instagram, and Threads992        with a more personalized approach so that people who want to993        see more political content in their feeds can.994995    We are also continually testing how we deliver personalized996        experiences and have conducted testing around civic content. As997        a result, we started treating civic content from people and998        Pages users follow on Facebook more like any other content in999        their feed, and we started ranking and showing people that1000        content based on explicit signals (for example, liking a piece1001        of content) and implicit signals (like viewing posts) that help1002        us predict what is meaningful to people. We are also1003        recommending more political content based on these personalized1004        signals and are expanding the options people have to control1005        how much of this content they see.10061007    These changes are an attempt to return to our ideals around free1008expression. That means being vigilant about the impact our policies and1009systems are having on people's ability to make their voices heard and1010having the humility to change our approach when we know we're getting1011things wrong.1012    We believe these changes are having a positive impact. As reflected1013in our Community Standards Enforcement Report for the first quarter of10142025, we saw a roughly 50 percent reduction in enforcement mistakes on1015our platforms in the United States from the last quarter of 2024 to the1016first quarter of 2025. During that same time period, the low prevalence1017of violating content largely remained unchanged for most problem areas.1018This improvement follows the earlier commitment we made to change our1019focus to proactively enforcing high-severity violations and enhancing1020our accuracy through system audits and additional signals, as described1021above.1022    Thank you, and I welcome your questions.10231024    The Chairman. Thank you. Mr. Creeley, you are recognized.10251026   STATEMENT OF WILL CREELEY, LEGAL DIRECTOR, FOUNDATION FOR1027                INDIVIDUAL RIGHTS AND EXPRESSION10281029    Mr. Creeley. Chairman Cruz, Ranking Member Cantwell, and1030honorable members of the Committee, good morning and thank you1031for the opportunity to testify today.1032    My name is Will Creely, and I am the Legal Director of1033FIRE, the Foundation for Individual Rights and Expression, a1034nonpartisan nonprofit dedicated to defending the rights of all1035Americans to free speech and free thought, the essential1036qualities of liberty. I have spent nearly 20 years defending1037the First Amendment rights of speakers from every point on the1038ideological spectrum.1039    At FIRE we have one rule, if it is protected, we will1040defend it. Typically, the censorship we fight is1041straightforward, the Government punishes the speaker for saying1042things the Government doesn't like. That is a classic First1043Amendment violation, a fastball down the middle. Unfortunately,1044that kind of textbook censorship isn't the only way Government1045actors silence disfavored or dissenting speech.1046    Far too often, Government officials from both sides of the1047partisan divide engage in jawboning. That is, they abuse the1048actual or perceived power of their office to threaten, bully,1049or coerce others into censoring speech. This indirect1050censorship violates the First Amendment just as surely as1051direct suppression.1052    This isn't new law. The First Amendment's prohibition1053against coerced censorship dates back decades to the Supreme1054Court's 1963 ruling in Bantam Books v. Sullivan. In that case,1055the court confronted a Rhode Island State Commission that sent1056threatening letters ``phrased virtually as orders'' to1057booksellers distributing ``objectionable titles'' with follow-1058up visits from police to ensure the message had been received.1059    The court held the commission's operation was ``in fact a1060scheme of state censorship effectuated by extra-legal1061sanctions.'' They acted as an agency not to advise but to1062suppress. And in this decade since, courts have consistently1063heeded Bantam Books' call to ``look through forms to the1064substance of censorship and to remain vigilant against both1065formal and informal schemes to silence speech.''1066    But Government officials regularly abuse their power to1067silence others, so the lesson of Bantam Books bears repeating.1068And in deciding National Rifle Association v. Vullo last year,1069the Supreme Court emphatically and unanimously reaffirmed it.1070In Vullo, New York State officials punished the NRA for its1071views on gun rights by threatening regulatory enforcement1072against insurance companies that did business with the group1073and offering leniency to those who stopped.1074    New York's backdoor censorship was successful and likely1075unlawful. This regulatory carrot and stick approach was1076designed to chill speech, and the court reiterated that ``a1077Government official cannot do indirectly what she is barred1078from doing directly.''1079    A Government official cannot coerce a private party to1080punish or suppress disfavored speech on her behalf. To be sure,1081the Government may speak for itself, and the public has an1082interest in hearing from it, but it may not wield that power to1083censor. As Judge Richard Posner put it, the Government is1084``entitled to what it wants to say, but only within limits.''1085    Under no circumstances may our public servants ``employ1086threats to squelch the free speech of private citizens.'' So,1087the law is clear, Government actors cannot silence a speaker by1088threatening, ``we can do this the way, or we can do this the1089hard way,'' as the Chairman of the Federal Communications1090Commission did last month.1091    Nevertheless, recent examples of jawboning abound against1092private broadcasters, private universities, private social1093media platforms, and more. The First Amendment does not abide1094mob tactics. Despite the clarity of the law, fighting back1095against jawboning is difficult.1096    Targeted speakers can't sue Federal officials for monetary1097damages for First Amendment violations, removing a powerful1098deterrent. And as a practical matter, informal censorship is1099often invisible to those silenced.1100    That is particularly true in the context of social media1101platforms, as demonstrated by another recent Supreme Court1102case, Murthy v. Missouri. Murthy involved coercive demands by1103Biden Administration officials to social media platforms about1104posts related to COVID-19, vaccines, elections, and other1105subjects, resulting in the suppression of speech the1106Administration opposed.1107    But the court held the plaintiffs lack standing to sue1108because the causal link between their deleted posts and the1109Administration's pressure wasn't sufficiently clear. Murthy1110illustrates a severe information disparity. Users whose speech1111is suppressed have no way to know if the Government actors put1112their thumb on the scale.1113    Only the Government and the platforms have that knowledge,1114and usually neither wants to share it. That is why FIRE1115authored model legislation that would require the Government to1116disclose communications between Federal agencies and social1117media companies regarding content published on its platform,1118with limited exceptions. Transparency is not enough.1119    Federal officials must be meaningfully deterred from1120jawboning and held accountable when they do. Jawboning betrays1121our national commitment to freedom of expression. Congress1122should take action to stop it. Thank you for your time. I1123welcome your questions.1124    [The prepared statement of Mr. Creeley follows:]11251126  Prepared Statement of Will Creeley, Legal Director, Foundation for1127                    Individual Rights and Expression1128    Chairman Cruz, Ranking Member Cantwell, and honorable members of1129the Committee,11301131    Good morning, and thank you for the opportunity to testify today.1132My name is Will Creeley, and I am the Legal Director of FIRE--the1133Foundation for Individual Rights and Expression, a nonpartisan1134nonprofit dedicated to defending the rights of all Americans to free1135speech and free thought, the essential qualities of liberty.1136    I've spent nearly twenty years defending the First Amendment rights1137of speakers from every point on the ideological spectrum. At FIRE, we1138have one rule: If speech is protected, we'll defend it.1139    Typically, the censorship we fight is straightforward: The1140government punishes a speaker for saying things the government doesn't1141like. That's a classic First Amendment violation, a fastball down the1142middle. Unfortunately, that kind of textbook censorship isn't the only1143way government actors silence disfavored or dissenting speech.1144    Far too often, government officials from both sides of the partisan1145divide engage in ``jawboning''--that is, they abuse the actual or1146perceived power of their office to threaten, bully, or coerce others1147into censoring speech. This indirect censorship violates the First1148Amendment just as surely as direct suppression.1149    This isn't new law. The First Amendment's prohibition against1150coerced censorship dates back decades, to the Supreme Court's 19631151ruling in Bantam Books v. Sullivan. In that case, the Court confronted1152a Rhode Island state commission that sent threatening letters,1153``phrased virtually as orders,'' to booksellers distributing1154``objectionable'' titles--with follow-up visits from police, to ensure1155the message had been received.1156    The Court held the commission's ``operation was in fact a scheme of1157state censorship effectuated by extra-legal sanctions; they acted as an1158agency not to advise but to suppress.'' And in the decades since,1159courts have consistently heeded Bantam Books' call to ``look through1160forms to the substance'' of censorship, and to remain vigilant against1161both formal and informal schemes to silence speech.1162    But government officials regularly abuse their power to silence1163others, so the lesson of Bantam Books bears repeating. And in deciding1164National Rifle Association of America v. Vullo last year, the Supreme1165Court unanimously and emphatically reaffirmed it.1166    In Vullo, New York State officials punished the NRA for its views1167on gun rights by threatening regulatory enforcement against insurance1168companies that did business with the group and offering leniency to1169those who stopped. New York's backdoor censorship was successful--and1170unlawful.1171    This regulatory carrot-and-stick approach was designed to chill1172speech, and the Court reiterated that ``a government official cannot do1173indirectly what she is barred from doing directly: A government1174official cannot coerce a private party to punish or suppress disfavored1175speech on her behalf.''1176    To be sure, the government may speak for itself, and the public has1177an interest in hearing from it. But it may not wield that power to1178censor. As Judge Richard Posner put it: The government is ``entitled to1179what it wants to say--but only within limits.'' Under no circumstances1180may our public servants ``employ threats to squelch the free speech of1181private citizens.''1182    So the law is clear: Government actors cannot silence a speaker by1183threatening ``we can do this the easy way or we can do this the hard1184way,'' as the chairman of the Federal Communications Commission did1185last month. Nevertheless, recent examples of jawboning abound: against1186private broadcasters, private universities, private social media1187platforms, and more. The First Amendment does not abide mob tactics.1188    Despite the clarity of the law, fighting back against jawboning is1189difficult. Targeted speakers can't sue Federal officials for monetary1190damages for First Amendment violations, removing a powerful deterrent.1191And as a practical matter, informal censorship is often invisible to1192those silenced.1193    That's particularly true in the context of social media platforms,1194as demonstrated by another recent Supreme Court case, Murthy v.1195Missouri.1196    Murthy involved coercive demands by Biden administration officials1197to social media platforms about posts related to Covid-19, vaccines,1198elections, and other subjects, resulting in the suppression of speech1199the administration opposed. But the Court held the plaintiffs lacked1200standing to sue, because the causal link between their deleted posts1201and the administration's pressure wasn't sufficiently clear.1202    Murthy illustrates a severe information disparity: Users whose1203speech is suppressed have no way to know if government actors put their1204thumb on the scale. Only the government and the platforms have that1205knowledge, and usually neither want to share it.1206    That's why FIRE authored model legislation that would require the1207government to disclose communications between Federal agencies and1208social media companies regarding content published on its platform,1209with limited exceptions. But transparency is not enough. Federal1210officials must be meaningfully deterred from jawboning, and held1211accountable when they do.1212    Jawboning betrays our national commitment to freedom of expression.1213Congress should take action to stop it.1214    Thank you for your time. I welcome your questions.12151216    The Chairman. Thank you. Mr. Feld, you are recognized.12171218    STATEMENT OF HAROLD FELD, SENIOR VICE PRESIDENT, PUBLIC1219                           KNOWLEDGE12201221    Mr. Feld. Chairman Cruz, Ranking Member Cantwell, thank you1222for inviting me to testify here today. At the heart of this1223hearing lies one of the most important questions for democracy,1224when does the normal functioning of a Federal agency cross the1225line into impermissible coercion? When does the bully pulpit1226become jawboning?1227    This clearly depends on context. It would be a shame if1228something happened to your business sounds very different1229coming from insurance broker than from a mobster. The Supreme1230Court has provided us with some guidance. For Government1231interaction to be considered an impermissible coercive act1232rather than a permissible act of persuasion, several conditions1233must exist.1234    First, it must be a Government contact, not interaction1235between private citizens or advocacy groups. Research by1236academics, calls by advocates to boycott companies or products1237can never under the First Amendment qualify as jawbone. To the1238contrary, such activities are political speech the First1239Amendment protects from Government retaliation. Second, the1240Government agency or official must have actual power to punish1241or reward the speaker.1242    And finally, there must be sufficient evidence that the1243Government agency or official has genuine intent to use that1244power. It is the uncertainty around the last two criteria that1245often make it difficult to tell when permissible persuasion and1246use of the bully pulpit cross the line. Unfortunately, whatever1247one thinks happened in the past, we live in a present that1248leaves no doubt that jawboning is occurring regularly.1249    And whatever one's political party, Members of Congress1250must join together to protect the fundamental freedom of speech1251to which our democracy depends. To illustrate my point, I want1252to contrast the first Trump Administration with today.1253President Trump is a master of the bully pulpit.1254    During his first Administration, he would complain about1255what he felt was one-sided news coverage or unfair mockery. He1256called for NBC to lose its license. But then FCC Chair Ajit Pai1257took no action on these threats. When asked, Pai would point to1258John F. Kennedy calling his FCC Chair, Newt Minow, to demand1259similar punishment of his critics.1260    Without saying so directly, Pai made it clear that1261Presidents have the bully pulpit, but agency heads must stay1262independent and follow the law. As we have seen, however, the1263current FCC Chair Brendan Carr feels differently.1264    Rather than act as a firewall between the President to the1265use of Government coercion, Carr has sacrificed the1266independence of the agency and repeatedly used the very real1267power of the FCC over licensees to meddle in content decisions1268that past FCC chairs of both parties have considered off1269limits. Most recently, he famously told Disney and its1270affiliates that they should remove Jimmy Kimmel, whom the1271President had repeatedly said through his social media account1272should be fired, ``the easy way or the hard way''.1273    Nor is this the only example where traditionally1274independent agencies such as the Federal Trade Commission have1275acted to punish those critical of the President or his1276policies. Indeed, it is not simply the acts of these agency1277heads, but the combination of multiple actions across multiple1278agencies that make the threat of Government retaliation very1279real.1280    Fortunately, Chairman Cruz has shown us what needs to be1281done to stop jawboning. Congress must act, as individuals and1282collectively, without regard to party or partisanship to draw1283clear lines in the sand. The first step is a willingness to1284call out such behavior. Chairman Cruz's rebuke of Chairman Carr1285for using language right out of Goodfellas is a model of1286standing for principle when called for.1287    Additionally, members must work to restore the independence1288of agencies and to set clear limits on the power of agencies to1289meddle in matters protected by the First Amendment.1290Unfortunately, the Supreme Court appears poised to eliminate1291the power of Congress to protect agency independence by1292preventing the President from firing Commission members.1293    But Congress can pass legislation to promote transparency1294and make it easier for those threatened to defend themselves in1295court. Such legislation, however, must be carefully crafted to1296avoid interfering with the legitimate contacts between1297Government and companies, or creating a disincentive to1298necessary oversight and enforcement.1299    Finally, Congress must address the concentration of1300ownership in both legacy media and digital platforms. The1301active effort by the Trump Administration to put Warner1302Brothers, owner of CNN, in the hands of proven friends while1303threatening to use the regulatory power of the United States to1304block bidders it dislikes illustrates the importance of1305diversified ownership to the marketplace of ideas.1306    The last few months have shown the importance of ownership1307limits on legacy media. Limits Chairman Carr has indicated he1308intends to relax or eliminate. Additionally, Congress should1309consider legislation to introduce competition to the digital1310platform phase.1311    Congress should consider cross-ownership limits that would1312prevent things like the acquisition of TikTok, one of the most1313important and influential digital platforms, by the largest1314owners of legacy media.1315    This concentration makes it easy for this Administration or1316any future Administration to monopolize the marketplace of1317ideas to the detriment of us all. Thank you, and I am happy to1318answer any questions.1319    [The prepared statement of Mr. Feld follows:]13201321       Prepared Statement of Harold Feld, Senior Vice President,1322                            Public Knowledge1323    Chairman Cruz and Ranking Member Cantwell, thank you for the1324opportunity to testify on such a critical topic. I have been an1325advocate for First Amendment freedoms--especially the right of the1326public to hear diverse views and production of local news critical to1327an informed democracy--for over 25 years. One of the most challenging1328areas of government policy is how to balance the important government1329interest in creating spaces for robust debate, ensuring the flow of1330necessary information, while simultaneously avoiding the twin dangers1331of government censorship or forced carriage of government propaganda.1332    At the beginning of the 21st Century, these concerns focused on the1333traditional gatekeepers of broadcast licensees, and preserving an open1334internet. More recently, the rise of social media and the concentration1335of our national discussions on a very few giant platforms have raised1336even more complicated questions. But the fundamental question remains1337the same. How can the government maintain the news and civic discourse1338necessary to sustain democracy without becoming the dictator of what1339people may say or hear?1340    Unfortunately, since President Trump was sworn into his second1341term, we have not had to worry about subtle intimidation or1342``regulation by raised eyebrow.'' We have seen an unambiguous pattern1343where the President has publicly boasted that he will take action1344against those who oppose his agenda or criticize him personally, and1345where the heads of theoretically independent agencies have rushed to1346make good on these threats. My personal expertise is with the Federal1347Communications Commission, and on this I will focus my testimony. But1348concerns about government censorship extend far beyond the FCC. I will1349touch briefly on the actions of the Federal Trade Commission and other1350Executive actions that reinforce the credibility of retaliation for1351expressing opinions contrary to those favored by the Administration.1352These actions contrast not only with the actions of the Biden1353administration but with the actions of the first Trump Administration.1354It is this pattern of public statements and regulator actions that1355makes threats real, and gives them coercive power.1356    I cannot stress too much how consolidation--particularly in the1357hands of those who have proven themselves loyal to the current1358Administration--amplifies the ability to control content on an1359unprecedented scale. Consider the recent statement in the New York Post1360that the President favors a takeover by Larry and David Ellison (who1361have been ``vetted'' by the changes they have made at CBS since it was1362acquired by Skydance) that any other bidders for Warner Bros ``are1363likely to face stiff hurdles from U.S. regulators.'' The Omnicom1364merger, approved by the FTC, created a behemoth that overshadows the1365entire advertising industry, making it harder for advertisers to1366control the placement of their ads. Social media remains highly1367concentrated, giving a handful of opaque algorithms control over the1368vast majority of online discussions and access to news. And without net1369neutrality, Internet providers may freely prioritize content that the1370administration favors and degrade content of its political opponents.1371    It is therefore critical for Congress to reassert itself as the1372protector of free speech and opponent of censorship, regardless of1373party. Chairman Cruz deserves credit for admonishing FCC Chairman Carr1374at a critical moment. If we want an open marketplace of ideas critical1375for democracy and self-governance, all members of Congress must act1376together to preserve free speech. We need laws that protect privacy so1377that people may say what they want without fear. We need laws that1378promote competition so that people have a choice in what they see,1379hear, and say. And we need to restore the principle of non-1380discrimination for broadband, a principle that has enjoyed bipartisan1381support even where parties remain split on the appropriate legal1382framework.1383          I. The ``Bully Pulpit'' v. Unconstitutional Coercion1384    President Theodore Roosevelt is credited with inventing the term1385``bully pulpit,'' meaning a conspicuous position to advocate an agenda.1386Certainly, since then, Presidents, members of Congress, and heads of1387agencies of both parties have used the bully pulpit to reflect their1388agendas and influence action. Indeed, it has long been regarded as an1389important function of elected officials to reflect the concerns and1390priorities of the people who elected them in speeches and in subsequent1391policy actions. That is not coercion, even if these statements have an1392influence on the actions of companies or individuals.1393    By contrast, the Supreme Court has consistently held--and scholars1394of the First Amendment broadly agree--that the government may not cross1395from pushing an agenda to compelling obedience and suppressing dissent.1396The power of the Executive Branch to prosecute--or even investigate--1397individuals, the power to deny economic benefits or, by contrast grant1398favors, gives the government tremendous coercive power. Often, the only1399way to tell the difference between the two is the impact. Do those who1400defy the President face more than social consequences, such as boycotts1401by the President's supporters? Are benefits or punishments explicitly1402tied to obedience or defiance, and if so, does the President carry1403through on these threats?1404    The Supreme Court recently illustrated the difference between1405persuasion and coercion through two cases.\1\ Taking these cases1406together, we find a clear set of criteria (albeit not always so simple1407to apply). For government statements to cross the line from persuasion1408to coercion, the individual speaking for the government must make a1409credible threat clearly based on expressing a disfavored viewpoint, or1410the refusal to express a viewpoint the government favors. The targeted1411individual or company must then either change its behavior or suffer1412consequences directly traceable to the behavior. The credibility of the1413threat is both a function of the ability of the official to impose the1414punishment (or dispense the reward) and the general understanding of1415the likelihood of the threat. This includes whether the agency in1416question is acting in a way contrary to its precedent on matters the1417agency does not generally opine.1418---------------------------------------------------------------------------1419    \1\ National Rifle Ass'n of Am. v. Vullo, 602 U.S. 175 (2024)1420(holding that the New York State Department of Financial Services had1421coerced financial institutions from providing services to the National1422Rifle Association) compare with Murthy v. Missouri, 603 U.S. 370 (2024)1423(holding that the government's communications with social media1424platforms did not cause a direct and redressable injury to the1425plaintiff).1426---------------------------------------------------------------------------1427    Thus, in Vullo, when the New York state official with relevant1428authority directly told a major insurer that it was much more likely to1429investigate companies providing insurance to the NRA, and issued1430official guidance reiterating this position, the Supreme Court found1431that this was an impermissible act of coercion against the NRA by1432depriving it of access to affordable insurance based on its views. By1433contrast, the Court in Murthy found that, even assuming the contacts1434between social media companies and the Executive branch took place,1435they did not constitute coercion. The Court found there was no obvious1436nexus between the Executive contacts and any reward or punishment for1437the social media companies and no obvious change in behavior in1438response to these contacts.\2\ Additionally, the Court recognized that1439the Administration had good reason to coordinate with social media1440companies with regard to news about the ongoing COVID pandemic and1441other matters potentially bearing on public safety.1442---------------------------------------------------------------------------1443    \2\ Murthy v. Missouri, 603 U.S. 370 (2024) at 59.1444---------------------------------------------------------------------------1445A. The FCC Has Crossed the Line to Unconstitutional Coercion and1446        Censorship1447    With this in mind, we must view the ongoing activities of the FCC1448under Chairman Brendan Carr with considerable alarm--especially when1449contrasted with the actions of Chairman Ajit Pai, who served as1450Chairman of the FCC in the first Trump Administration. During the First1451Trump administration, President Trump would frequently denounce his1452critics on television--particularly those on Saturday Night Live who1453mocked him. President Trump would post that NBC should ``lose its1454license'' for its unfair coverage. Chairman Pai, however, took no1455action nor suggested any action to make good on these threats. To the1456contrary, when asked about the President's threats, Pai would refer to1457similar complaints by President Kennedy to then-FCC Chair Newt Minnow.1458While not saying so directly, Chairman Pai clearly implied that the1459President should be free to make his anger known (an example of the1460bully pulpit), while the Chairman of the FCC should remain independent.1461    But things have been very different this time around. Throughout1462his campaign, President Trump promised to punish his enemies when he1463returned to office. He also promised to use the power of government to1464punish those who stood in the way of his agenda. Brendan Carr, who as1465Chair of the FCC could make these threats credible, began to take1466actions that seemed very much in line with these threats. One of his1467first acts as Chair was to reinstate complaints for ``news distortion''1468against ABC, NBC, and CBS on the basis of what were alleged to be1469unfairly favorable coverage of Vice President Kamala Harris during the14702024 campaign that Chairwoman Jessica Rosenworcel had previously1471dismissed.1472    These actions were noteworthy for multiple reasons. For one thing,1473none of the parties filed a Petition for Reconsideration or any other1474request to reinstate the complaints.\3\ Furthermore, news distortion1475complaints are extremely difficult to prove. The First Amendment1476guarantees to broadcasters the right to edit their broadcasts.1477Additionally, Section 326 of the Communications Act expressly prohibits1478the FCC from engaging in censorship. This made reinstating the1479complaints after an opinion in line with Commission precedent1480extraordinarily peculiar. Finally, even taking the allegations on their1481face, the matters were moot or quickly became moot. In the case of the1482complaint against NBC for violating the ``equal time'' rule,\4\ NBC1483reported that it had offered the Trump campaign comparable time and1484that the Trump campaign had accepted the offer. In the accusation that1485CBS had edited an interview with Vice President Harris to make her1486appear more favorably, the complainants requested as relief that the1487FCC release the unedited transcript, which the FCC subsequently did.\5\1488---------------------------------------------------------------------------1489    \3\ By contrast, the Media Bureau simultaneously dismissed a1490complaint against Fox29 in Philadelphia for violations of the1491Commission's character policy based on the claims that the 20201492election was stolen by manipulating an electronic voting machine1493provider--a claim which Fox News subsequently settled. Chairman Carr1494declined to reinstate that complaint.1495    \4\ 47 U.S.C. Sec. 312(a)(7) and Sec. 315(a) generally require1496broadcasters to provide equal time, at similar rates, to qualified1497Federal candidates.1498    \5\ Center for American Rights, In re Complaint Against WCBS-TV, at14995 (filed October. 16, 2024), https://drive.google.com/file/d/1kBqZo-150010xBLE0Y1dhvBpzZnvcRUvH0H4/view.1501---------------------------------------------------------------------------1502    Carr has often echoed criticism made by President Trump that media1503coverage is too critical of Trump and his agenda, favors Democrats and1504liberal viewpoints, and that stations that do not alter their coverage1505may be guilty of a ``pattern of news distortion'' and may be violating1506their public interest obligations. Carr has announced various1507investigations against outlets Trump has expressly criticized, such as1508Comcast/NBC Universal and NPR and PBS. Carr has expressly targeted for1509investigation a broadcast station that covered an ICE raid in a manner1510unflattering to ICE. All of these fall within the pattern the court1511identified in Vullo as efforts at unconstitutional coercion.1512    By contrast, Chairman Carr has used his power as Chair to weigh in1513favorably for licensees who explicitly back the President and have1514owners known to be President Trump's allies. Networks and affiliates1515regularly negotiate the terms by which the networks will provide1516affiliates with programming. Chair has written to Comcast/NBCU that it1517is investigating its treatment of its affiliates to influence these1518negotiations,\6\ and generally voiced concern over the treatment of1519networks over their affiliates, when the largest group owners of1520affiliates--Nexstar and Sinclair--are owned by allies of President1521Trump and program accordingly.1522---------------------------------------------------------------------------1523    \6\ Letter from Brendan Carr, Chairman, Fed. Commc'ns Comm'n, to1524Brian Roberts, CEO, Comcast Corp. (Feb. 11, 2025), https://www.fcc.gov/1525sites/default/files/Chairman-Carr-Letter%152620to-Comcast-02112025.pdf.1527---------------------------------------------------------------------------15281. Merger Conditions and Targeting Talk Show Hosts1529    The most significant instances of censorship have come from1530Skydance's acquisition of Paramount and the subsequent ``easy way or1531the hard way'' demand to terminate Jimmy Kimmel. While citizen action1532ultimately forced Disney to reverse its decision to suspend broadcast1533of Jimmy Kimmel, the incident still raises very troubling concerns over1534Chairman Carr's willingness to use the power of the FCC to reward and1535punish to directly control content. Additionally, these incidents show1536the impact of media consolidation on the power of censorship and why1537Congress must take immediate steps to prevent further consolidation.1538    Paramount/Skydance. President Trump has been extremely critical of1539CBS, particularly 60 Minutes, and what he perceives as the persistent1540negative coverage of his Presidency. Trump has also been highly1541critical of Late Night talk show host Stephen Colbert, who has been one1542of the President's most vocal and persistent critics. Because the sale1543of Paramount to Skydance required FCC approval, it created the1544opportunity for Chairman Carr to effectively demand changes to1545Paramount's programming and commitments to provide more favorable1546coverage of President Trump and his agenda.1547    The Skydance acquisition did not present a traditional problem of1548media consolidation. Although Skydance owned some entertainment1549properties, it did not own any broadcast licenses or cable channels.1550While the merger might have arguably raised some concerns over1551concentration in media generally, it did not raise concerns that the1552FCC has considered in previous transactions. The transaction should1553therefore have gone through in fairly short order based on traditional1554FCC analysis.1555    Instead, the transaction lingered for months. While Chairman Carr1556did not (at least not publicly) make specific demands, Paramount1557apparently understood that it would facilitate regulatory approval if1558it took steps to address President Trump's complaints that CBS had1559treated Trump badly and that it would avoid ``unbalanced'' and1560``biased'' coverage going forward. First, CBS settled a pending lawsuit1561filed by Trump relating to its editing of the ``60 Minutes'' interview1562with Kamala Harris. Although CBS admitted no wrongdoing and maintained1563that the settlement was simply to resolve the litigation, others argued1564that the settlement was motivated so that the acquisition could move1565forward.\7\ Stephen Colbert, host of The Late Show, denounced the1566settlement as a ``big fat bribe.''1567---------------------------------------------------------------------------1568    \7\ See David Folkenflik, ``CBS is the Latest News Giant to Bend to1569Trump's Power,'' All Things Considered (July 2, 2025), https://1570www.npr.org/2025/07/02/nx-s1-5454790/cbs-settlement-trump-60-minutes-1571harris-interview-analysis; Kinsey Crowley, Kathryn Palmer, and Mike1572Snyder, ``Why is Trump Suing CBS? What to Know About the Complaint1573Dating Back to Kamala Harris Interview,'' USA Today (May 20, 2025),1574https://www.usatoday.com/story/news/politics/2025/05/20/trump-cbs-1575lawsuit-settlement-explained/837420-007/.1576---------------------------------------------------------------------------1577    Almost immediately after Colbert's statement, CBS canceled the Late1578Show, effective at the end of the season in May 2026. While Trump and1579Carr both publicly celebrated the cancellation, both also denied that1580they had been responsible for CBS' decision.\8\ Finally, Skydance CEO1581David Ellison filed in the merger docket a letter to Chairman Carr in1582the FCC record committing to numerous changes to ``reflect the varied1583ideological perspective of American viewers.'' \9\ When asked whether1584these were necessary for the acquisition to get approval, Carr1585demurred, stating that Paramount canceled Colbert for financial reasons1586and that the FCC would continue to review the Paramount/Skydance deal1587at its own pace.The fact that the acquisition cleared the FCC mere1588hours later, however, undercut these assurances. And although Trump1589denied direct involvement in Colbert's cancellation, he posted on Truth1590Social that ``I absolutely love that Colbert got fired . . . I hear1591Jimmy Kimmel is next.'' \10\1592---------------------------------------------------------------------------1593    \8\ Domenick Mastrangelo, ``Trump Says He Didn't Get Colbert1594Cancelled,'' The Hill (July 29, 2025), https://thehill.com/homenews/1595administration/5426557-trump-cbs-colbert-cancellation/.1596    \9\ Letter of Mathew A. Brill to Marlene H. Dortch, Skydance Media1597and Paramount Global Application for Transfer of Control of Licenses,1598MB Docket No. 24-275 (filed July 17, 2025), https://www.fcc.gov/ecfs/1599document/1071757519667/1.1600    \10\ Truth Social Post of Donald Trump, https://truthsocial.com/1601@realDonaldTrump/posts/114874422468516376.1602---------------------------------------------------------------------------1603    Carr soon seized the opportunity to fulfill President Trump's1604``prediction.'' Following the tragic assassination of Charlie Kirk,1605Jimmy Kimmel used his television show to criticize those in the MAGA1606movement, blaming liberals generally for Kirk's death.\11\ Chairman1607Carr moved from ``regulation by raised eyebrow'' to full-on public1608threat. ``This is a very, very serious issue right now for Disney,''1609Chairman Carr told conservative podcaster Benny Johnson. In language1610which Chairman Cruz would rightly criticize as ``right out of1611`Goodfellas,' Chairman Carr continued: ``We can do this the easy way or1612the hard way. These companies can find ways to take action on Kimmel,1613or there is going to be additional work for the FCC ahead.'' Within1614hours, Nexstar, Tegna, and Sinclair--the three largest broadcast group1615owners controlling access to well over 25 percent of the country--1616announced they would preempt Kimmel. Soon after that, ABC announced it1617had suspended Kimmel ``indefinitely.'' Carr applauded this1618``independent decision'' of the affiliates as an example of how local1619affiliates serve their local communities.1620---------------------------------------------------------------------------1621    \11\ At the time, the assassin's motive (and potential political1622affiliation) was unclear.1623---------------------------------------------------------------------------1624    Public outrage was swift. Critics of Carr's statements noted that1625Nextstar and Tegna had already announced plans to merge--a deal that1626would not only require FCC approval, but actually exceeded FCC1627ownership limits and therefore would require either waiver or1628modification of the ownership rules (something Carr had previously1629indicated he favored). Additionally, the fact that three CEOs could1630dictate what the country would see brought home the way in which media1631consolidation has made censorship easier. Customers of Disney+ and Hulu1632(majority owned by Disney) began canceling their subscriptions in1633protest. Even Disney actors joined in the call to boycott Disney in1634protest.\12\ Ultimately, after Disney lost 3 million streaming1635customers,\13\ the company brought back Kimmel. Although Nexstar, with1636Carr's encouragement, preempted Kimmel's return broadcast, it and the1637other two group owners ultimately relented, and Kimmel is now available1638again on all ABC affiliates.1639---------------------------------------------------------------------------1640    \12\ Jason Ma, ``Even Disney Stars Are Joining Calls to Boycott the1641Media Giant After ABC Suspended Jimmy Kimmel's Show, Fortune (Sept. 20,16422025), https://www.yahoo.com/entertainment/celebrity/articles/even-1643disney-stars-joining-calls-224642755.html.1644    \13\ Lindsay Kornick, ``Disney+ Lost Nearly 3 Million Subscribers1645After Jimmy Kimmel Suspension: Report,'' Fox Business (Oct. 21, 2025),1646https://www.foxbusiness.com/media/disney-lost-nearly-three-million-1647subscribers-after-jimmy-kimmel-susp ension-report.1648---------------------------------------------------------------------------1649B. This is Not How the Public Interest Works1650    Carr has consistently invoked the public interest standard and1651traditional media policies of promoting localism and diversity of views1652as justification for his investigations and threats. But Carr's actions1653contradict the FCC's traditional policies that carefully avoid1654influencing content. Especially since the end of the Fairness Doctrine1655in the 1980s, the FCC has relied on structural limits such as ownership1656limits to provide for a diversity of owners to ensure diverse1657viewpoints. Even when the FCC did have rules designed to ensure local1658programming, they were either structural rules (such as the Financial1659Syndication rules, the Prime Time Access Rules, or the Local Studio1660Rule) \14\ or explicitly neutral, such as the Fairness Doctrine.\15\1661Most of these rules were eliminated by the FCC or the courts during the16621980s and 1990s, leaving ownership limits as the one mechanism for1663ensuring viewpoint diversity in broadcasting.1664---------------------------------------------------------------------------1665    \14\ The Financial Syndication Rule encouraged independent1666programming by prohibiting networks from owning the programming they1667syndicated. The Prime Time Access Rule required that local stations air1668at least one hour of non-network programming at the beginning of prime1669time and at the end of prime time. The Local Studio rule required local1670stations to maintain a local studio for the production of independent1671programming.1672    \15\ The Fairness Doctrine was not a single rule, but multiple1673rules requiring licensees to cover issuance of importance to their1674local communities and to allow opportunities for those with opposing1675opinions to respond.1676---------------------------------------------------------------------------1677    Indeed, it is idle to speak of localism when decisions about access1678to broadcast programming are made for dozens of markets by a single1679corporate CEO. It is no more localism for a programming decision to1680come from a Federal building in Washington, D.C. than for the decision1681to come from Sinclair's corporate headquarters an hour north in1682Baltimore. Carr's insistence that his dictates are merely appeals to1683traditional FCC concerns for localism and the broader public interest1684cannot stand against a repeated pattern of attacking those that1685President Trump publicly attacks, while rewarding those the President1686favors. It cannot square with the traditional ways in which the FCC has1687addressed the policies Carr cites as justification for his actions. It1688turns localism on its head to allow further consolidation. By contrast,1689it makes control of content easier when an ever smaller number of ever1690larger companies controls what the public sees and hears.1691   II. Other Agencies Are Acting To Create an Engine of Censorship,1692       Feeding the Credibility of a General Threat to Free Speech1693    Chairman Carr is not unique as the head of an independent agency1694now openly working to further the President's agenda on punishing1695critics and rewarding those who demonstrate loyalty. To the contrary,1696the President has enlisted every agency at his command to fulfill his1697campaign promise of punishing his enemies. This, in turn, gives every1698individual threat of retaliation greater credibility and provides1699important context. Statements that in Trump's first administration1700could be dismissed as use of the bully pulpit to express displeasure1701must now be considered as made with coercive intent.1702    It is important to keep this in mind when considering whether the1703same statements from the Biden Administration, or even the previous1704Trump Administration, are coercive jawboning. The statement, ``Nice1705place you got here, it would be a shame if something happened to it,''1706sounds entirely different when said by an insurance salesman1707A. The FTC Acts Beyond Its Statutory Authority17081. Media Matters lawsuit1709    Media Matters for America, a liberal watchdog group, is a1710particular thorn in Elon Musk's side. So much so that Musk sued Media1711Matters in November 2023, claiming the group had manufactured a report1712on advertising placed next to extremist content on Musk's X,1713effectively driving away advertisers from the platform. While this case1714is ongoing, one of the first initiatives undertaken by FTC Chair Andrew1715Ferguson was to open an investigation into Media Matters to determine1716whether the watchdog group indeed organized an advertiser boycott of1717social media platforms, like X. It is quite the coincidence the FTC1718that, among its first priorities, target the very same group that1719plagues Elon Musk, an avid supporter of Donald Trump who spent over1720$290 million to support Trump's election.\16\1721---------------------------------------------------------------------------1722    \16\ Tim Reid and Jason Lange, Musk spent over a quarter of a1723billion dollars to help elect Trump, Reuters (December 6, 2024),1724https://www.reuters.com/world/us/musk-spent-over-quarter-billion-1725dollars-help-elect-trump-2024-12-06/.1726---------------------------------------------------------------------------1727    Believing its First Amendment rights are under threat by the FTC,1728Media Matters filed suit in Federal court on June 23, 2025, alleging1729that the FTC is retaliating against the organization for its reporting1730on Elon Musk and X. On August 15, 2025, Judge Sparkle Sooknanan granted1731a preliminary injunction enjoining the FTC from enforcing its civil1732investigative demand. The judge determined that Media Matters engaged1733in quintessential First Amendment activity when it published the1734research regarding Musk's platform and found that the organization was1735likely to succeed on its First Amendment retaliation claim. Part of the1736justification pulls from social media posts from chairman Ferguson's1737own staffers expressing their disdain for Media Matters' watchdog1738efforts, including FTC Senior Policy Advisor Jon Schweppe, who wrote,1739``Media Matters = scum of the earth.''\17\ The timing of the1740investigation, combined with clear animus toward Media Matters by1741several of Ferguson's direct reports, was sufficient for the court to1742determine that the civil investigative demand to Media Matters was1743retaliatory, with the intention of stifling the watchdog's capacity to1744track and report on falsehoods in conservative media.1745---------------------------------------------------------------------------1746    \17\ Jon Schweppe @JohnSchweppe, (2023, November 30), [X], https://1747x.com/JonSchweppe/status/1730279114748166522.1748---------------------------------------------------------------------------1749    Judge Sooknanan wrote in the August ruling, ``It should alarm all1750Americans when the Government retaliates against individuals or1751organizations for engaging in constitutionally protected public debate.1752And that alarm should ring even louder when the Government retaliates1753against those engaged in newsgathering and reporting.'' And on October175423, a U.S. appeals court declined to reinstate the FTC's probe into1755Media Matters.\18\1756---------------------------------------------------------------------------1757    \18\ Wendy Davis, Court Preserves Block on FTC Probe of Media1758Matters, Media Daily News, (October 23, 2025), https://1759www.mediapost.com/publications/article/410132/court-preserves-block-on-1760ftc-probe-of-media-matter.html.1761---------------------------------------------------------------------------17622. Omnicom Merger1763    The FTC's consent order for the Omnicom-Interpublic merger1764represents another instance of the agency acting beyond its statutory1765authority to advance a political agenda rather than address genuine1766competitive concerns. FTC Chair Andrew Ferguson claimed the order was1767necessary to prevent the merged advertising agency from ``boycotting''1768digital platforms based on political content. However, this1769fundamentally misunderstands how media buying actually works.1770    Advertisers themselves make placement decisions based on their own1771brand safety standards, not their agencies. The industry's standard1772contracts make this explicit, with advertisers maintaining written1773control over what contexts are ``safe and protective'' of their brands.1774    Despite Chair Ferguson's rhetoric, the consent order itself1775preserves advertisers' ability to choose ad placements, which1776undermines the order's stated impact while potentially violating1777Omnicom's rights of association and independent refusals to deal.\19\1778The real burden falls on advertisers, who must now take on more1779responsibility for researching and preparing exclusion lists. As one1780trade publication put it, ``If you don't want your stuff to be shown1781against Nazis, you're going to have to figure out how, and instruct1782Omnicom accordingly.'' The order also requires Omnicom to report1783exclusion lists to the FTC for 4 years, potentially exposing clients to1784political attacks, even though political ideology isn't the basis for1785these lists.1786---------------------------------------------------------------------------1787    \19\ Lisa Macpherson and Elise Phillips, Update: The FTC's Consent1788Order in the Omnicom Ad Agency Merger Misses the Point, Public1789Knowledge, (June 27, 2025), https://publicknow1790ledge.org/update-the-ftcs-consent-order-in-the-omnicom-ad-agency-1791merger-misses-the-point/ (The FTC used the merger to effectuate1792information-control goals on behalf of the Trump administration, rather1793than addressing real monopolistic practices in online advertising that1794created the opacity that requires advertisers to create brand safety1795requirements).1796---------------------------------------------------------------------------1797    Most troubling is what the consent order ignores. The FTC's own1798complaint acknowledged serious competitive concerns about coordination1799that could harm consumers by reducing competition, lowering product1800quality, and slowing innovation. Industry experts warned the merger1801could ``limit contestability or stifle innovation in the market'' and1802impact pricing, ad placement, sponsorships, commission rates, media1803discounts, and more. Yet the consent order addresses none of these1804actual competitive harms. Instead, it focuses narrowly on speculative1805claims about boycotting politically controversial publishers, relying1806on reports from partisan political actors like Representative Jim1807Jordan and complaints from Elon Musk rather than robust economic1808analysis.1809    The narrow scope suggests the FTC prioritized the political impact1810of decreased revenue to publishers hosting Trump administration-1811friendly content over genuine concerns about diminished competition1812affecting advertisers, consumers, or workers. This pattern aligns with1813the FTC's recent use of antitrust concepts to further information1814control goals on behalf of the Trump administration, rather than1815addressing real monopolistic practices in online advertising that1816created the opacity requiring brand safety measures in the first place.1817B. Trump Administration's Pressure Campaign Against Universities and1818        Students to Force ``Viewpoint'' Changes is a Violation of the1819        First Amendment1820    In April this year, the Trump administration froze $2.2 billion in1821grants and $60 million in contracts to Harvard University after the1822school rejected Federal demands to make governance and leadership1823reforms and conduct an audit of viewpoint diversity of students and1824employees.\20\ Because Harvard asserted its right to academic1825independence and rejected Federal demands (while reiterating its1826efforts to improve its support and protection of Jewish students), in1827May 2025, the Trump administration cut another $450 million in Federal1828grants to Harvard, on top of the $2.2 billion already slashed, citing1829antisemitism concerns. In April, the American Association of University1830Professors (AAUP) and Harvard University filed a lawsuit seeking to1831block the Trump administration from demanding that Harvard University1832restrict speech and restructure its core operations or else face the1833cancellation of all $8.7 billion in Federal funding.1834---------------------------------------------------------------------------1835    \20\ Alvin Powell, Trump administration freezes $2.2 billion in1836grants to Harvard, Harvard School of Public Health, (April 15, 2025),1837https://hsph.harvard.edu/news/trump-administration-freezes-2-2-billion-1838in-grants-to-harvard/.1839---------------------------------------------------------------------------1840    On September 3rd, U.S. District Judge Allison Burroughs found that1841the Trump administration's actions violated Harvard and Harvard1842faculty's First Amendment Rights, finding that ``the government-1843initiated onslaught against Harvard was much more about promoting a1844governmental orthodoxy in violation of the First Amendment than about1845anything else, including fighting antisemitism.'' \21\ The government1846has the authority to decide which research projects it wants to fund;1847however, it cannot withdraw or deny funding based on the funding1848recipient's speech that is unrelated to the research. Moreover, if1849funding is terminated due to alleged violations of Title VI, as the1850Trump administration claimed, the government is legally required to1851follow a specific investigatory process. This process includes holding1852hearings and providing the opportunity to address the alleged1853violations.1854---------------------------------------------------------------------------1855    \21\ President & Fellows of Harvard Coll. v. U.S. Dep't of Health &1856Hum. Servs., No. 25-cv-11048-ADB, 2025 WL (D. Mass. Sept. 3, 2025),1857https://storage.courtlistener.com/recap/gov.uscourts.mad.283315/1858gov.uscourts.mad.283315.141.0_1.pdf.1859---------------------------------------------------------------------------1860    Drawing broad conclusions without conducting a proper investigation1861and withdrawing funding based solely on anecdotes or biases constitutes1862a violation of the First Amendment. And in the end, the Trump1863administration failed to explain how stripping billions in Federal1864funding would protect and improve the experience of Jewish students,1865especially considering many of those students rely on Federal funding1866for their own academic pursuits.1867    The AAUP also filed a lawsuit to block the Trump administration1868from carrying out arrests, detentions, and deportations of noncitizen1869students and faculty members based on their speech content. This1870lawsuit arose from the abrupt arrest and detention of students like1871Rumesysa Ozturk, who was targeted for writing an opinion article for1872the Tufts University newspaper that called for the university to divest1873from Israel. The opinion piece did not call for violence and did not1874glorify terrorists. It was written under the assumption that Ms. Ozturk1875could enjoy the freedom of speech guaranteed to individuals in the1876United States. However, for expressing her views, she was abducted from1877the streets of a Boston suburb and detained without due process.\22\1878---------------------------------------------------------------------------1879    \22\ Sergio Martinez-Beltran, Federal court rules Rumeysa Ozturk1880must be transferred to detention in Vermont, NPR, (May 7, 2025),1881https://www.npr.org/2025/05/07/g-s1-64726/federal-court-rules-student-1882rumeysa-ozturk-detention-vermont-challenge.1883---------------------------------------------------------------------------1884    The abduction and detention of students due to their pro-1885Palestinian activism have created a climate of fear and suppression on1886university campuses, effectively silencing legitimate political dissent1887that falls beyond the Trump administration's orthodoxy. On September188830, a Federal court ruled that the Trump administration's crackdown on1889students for their speech violates the First Amendment. The court found1890that ``Secretaries Noem and Rubio are engaged in a mode of enforcement1891leading to detaining, deporting, and revoking noncitizens' visas solely1892on the basis of political speech, with the intent of chilling such1893speech and that of others similarly situated. Such conduct is not only1894unconstitutional, but also almost unknown in our constitutional1895tradition.''1896    Despite these losses, the Trump administration has not let up in1897forcing University leaders to eschew First Amendment-protected academic1898freedom and come to heel to the President's ideological agenda. The1899White House has approached universities to agree to a ``compact'' in1900exchange for preferential treatment for research grants.\23\ Curiously,1901this suggests the government would select grant recipients not on1902merit--a criterion the White House has championed in its expansive1903efforts to dismantle Diversity, Equity, and Inclusion initiatives--but1904rather by a university's willingness to comply with White House-aligned1905ideological requirements. It is difficult to see how the compact is1906aligned with academic freedom and freedom of expression--a perspective1907shared by nearly all universities that received the compact.\24\1908---------------------------------------------------------------------------1909    \23\ Riana Mahtani, Universities join Penn in rejecting White House1910compact as consequences for refusal remain unclear, The Daily1911Pennsylvanian, (October 22, 2025), https://www.thedp.com/article/2025/191210/penn-white-house-compact-update-marc-rowan.1913    \24\ April Rubin, Colleges turn down Trump's ``compact'' agreement1914offer, Axios, (October 22, 2025), https://www.axios.com/2025/10/22/1915trump-universities-compact-agreement-rejected.1916---------------------------------------------------------------------------1917                 III. Some Things Cannot Be Censorship1918    Some things can be questionable whether they constitute unfair1919coercion, an exercise of the bully pulpit, or something in between. But1920some things are never censorship. First, we have long recognized the1921important responsibility of the government to protect people from1922harassment, from fraud, and from other forms of false, misleading or1923weaponized information. Section 223 (47 U.S.C. Sec. 223) of the1924Communications Act has prohibited harassment by telephone since 1968.1925In May of this year, Congress recognized the harm done by non-1926consensual intimate visual depictions (commonly referred to as1927``revenge porn'' or ``deep fake porn'') and required covered platforms1928to develop notice-and-takedown procedures for those harmed. (47 U.S.C.1929Sec. 223a) Laws criminalize election fraud, financial fraud, and other1930kinds of harmful and deceptive statements. We do not consider this1931censorship. In our increasingly complex world, good governance demands1932that Executive agencies and Congress respond to the emergence of new1933scams and threats.1934    Government officials, including election officials, have a1935responsibility to protect their constituents and ensure accurate1936information reaches the public, particularly regarding matters of1937public health, national security, and election integrity. When election1938officials identify objectively false information about polling1939locations, voting procedures, or ballot deadlines that could interfere1940with citizens' ability to vote, facilitating communication with1941platforms that may be inadvertently hosting such content serves a1942legitimate governmental function. This is information-sharing, not1943coercion. This includes when government entities work with university1944researchers to identify and assess how, for example, influence1945operations conducted by foreign adversaries may prevent Americans from1946participating in an election.1947    But the threats do not end there. We know that foreign governments1948seek to influence and disrupt us with bot armies spreading divisive and1949false information. Foreign terrorists use social media platforms to1950radicalize and recruit vulnerable individuals. Cybersecurity experts1951warn us that we must share information to identify these threats to1952protect ourselves, and that social media platforms must constantly1953engage in a cat-and-mouse game of content moderation to meet these1954threats. This sort of information sharing also cannot constitute1955censorship.1956    Government entities therefore can, even should, engage with1957platforms to share threat intelligence, coordinate on cybersecurity1958concerns, address foreign interference, and communicate about public1959health emergencies. These are standard government-industry coordination1960activities similar to information-sharing on terrorism and other1961threats. The key question is not whether the government communicates1962with platforms, but whether that communication crosses the line into1963coercion through threats of regulatory action, license revocation, or1964other forms of retaliation.1965    Platforms themselves have independent business and reputational1966reasons to address harmful content, including maintaining user trust1967and safety, protecting advertiser relationships, avoiding being vectors1968for foreign interference, and responding to international regulatory1969pressure. When platforms act on their own policies based on their own1970interests, this is not government censorship--it is private editorial1971decision-making protected by the First Amendment.1972A. Academic Research and Fact Checking is not Censorship1973    Academic institutions and researchers engaging in studies of online1974falsehoods, propaganda, and content moderation are exercising their own1975First Amendment rights, not participating in government censorship.1976When university researchers identify and document patterns of harmful1977content on social media platforms, this constitutes protected speech1978and legitimate scholarly inquiry. It is important for government1979agencies, as well as members of Congress, to consult this research to1980understand the nature of emerging threats. To the extent others feel1981that this research is biased or politically motivated, they have their1982own First Amendment right to critique this work.1983    The Election Integrity Partnership (EIP), which included Stanford1984Internet Observatory and the University of Washington Center for an1985Informed Public, has been incorrectly characterized by Senator Cruz's1986report on CISA as part of a government censorship apparatus. In1987reality, EIP was formed in 2020 under the Trump administration to help1988identify election-related misinformation that could interfere with1989voting. Researcher Renee DiResta of Stanford Internet Observatory1990explains in a guest essay to the New York Times, ``teams of student1991analysts identified social media posts that were potentially misleading1992the public about voting procedures or that tried to delegitimize the1993outcome of an election. Sometimes, a nonprofit clearinghouse that1994included state and local election officials shared with us posts that1995concerned them. In some cases, if a post we examined appeared to be1996going viral and appeared to violate a social media platform's election1997policies, we let the companies know. Most of the time, the platforms1998took no action; when they did act, it was primarily to label the post1999as disputed or to attach a fact check.'' \25\2000---------------------------------------------------------------------------2001    \25\ Renee DiResta, What Happened to Stanford Spells Trouble for2002the Election, New York Times, (June 25, 2024), https://www.nytimes.com/20032024/06/25/opinion/stanford-disinfor2004mation-election-jordan-twitter.html.2005---------------------------------------------------------------------------2006    There is no mechanism by which university researchers could coerce2007major technology companies as proxies for government officials.2008Academic researchers possess no regulatory authority, cannot levy2009fines, and cannot threaten business licenses or antitrust action.2010Platforms are under no obligation to act on academic research findings.2011The suggestion that platforms felt coerced by academic institutions2012fundamentally misunderstands the power dynamics at play. In fact,2013platforms welcomed assistance from researchers in identifying content2014that violated their own policies during unprecedented challenges2015related to election misinformation and public health crises, precisely2016because content moderation at scale is extraordinarily difficult.2017    When researchers publish findings on disinformation campaigns,2018document harassment patterns, or analyze content moderation practices,2019they contribute to public knowledge and democratic discourse. Attempts2020to characterize such research as censorship or government coercion not2021only misrepresent the nature of academic inquiry but also threaten to2022chill legitimate scholarship on critical issues affecting our2023democracy. The recent pattern of expensive lawsuits against academic2024institutions and civil society organizations focused on understanding2025and mitigating disinformation has already caused many to lose funding2026and talent, undermining our collective ability to understand and2027address these challenges.2028B. Calls for Legislation Are Not Censorship2029    It is the duty of elected officials to pass laws that serve the2030public interest--and to change laws that do not function as intended.2031Accordingly, it cannot be censorship to tell companies that if they2032refuse to change their ways, Congress will take action. Indeed, as an2033advocate, I often wish that Congress would step up and pass appropriate2034legislation rather than rely on industry self-regulation. But threats2035to introduce legislation fail as coercion for another reason. Neither2036the President nor a member of Congress has the power to unilaterally2037pass legislation. As every member of this Committee knows, passing2038legislation requires persuading a majority of members to pass a law,2039and for the President to sign it.2040    Calls for legislation in response to the actions of platforms, or2041their failure to act, are part of this process or persuasion. They are2042not censorship, but the robust debate on which our democracy depends.2043To the extent platforms act to dissuade Congress from acting, they are2044likewise engaged in the democratic process.2045    IV. Congress has a Duty To Demand Transparency And Oversight Of2046            Independent Agencies Regulating Speech Platforms2047    Platforms are more than happy to eschew blame for their moderation2048mistakes by scapegoating the government. We can acknowledge that2049platform policies developed during the first Trump administration and2050in the Biden administration, in part in response to U.S. government2051public health and election security priorities, resulted in some poor2052content decisions. But in the end, such were content moderation2053decisions, not acts of censorship. Importantly, we must distinguish2054between mistakes in what constitutes misinformation or disinformation2055as opposed to efforts to suppress entire viewpoints--especially when2056the definitions employed are vague and therefore subject to abuse.2057    For example, in the Executive Order ``Countering Domestic Terrorism2058and Organized Political Violence'', the White House has given explicit2059permission to go after organizations and individuals' speech that, in2060the Trump administration, represent ``anti-Americanism, anti-2061Capitalism, and anti-Christianity.'' \26\ To be clear, criticizing2062America, disliking capitalism, and disparaging Christianity may be2063viewed by many as deplorable, but it is not illegal. In fact, it is2064precisely protected by the First Amendment. More to the point, it is2065not entirely clear what these terms mean. Is an argument that Google2066has outsized marketpower ``anti-Capitalism''? Is a Protestant critique2067of Catholic dogma ``anti-Christianity?'' This sort of vagueness has2068been roundly condemned by the Supreme Court in multiple First Amendment2069decisions as creating a chilling effect on free speech.2070---------------------------------------------------------------------------2071    \26\ Presidential Memoranda, Countering Domestic Terrorism and2072Organized Political Violence, White House, (September 25, 2025),2073https://www.whitehouse.gov/presidential-actions/2025/09/countering-2074domestic-terrorism-and-organized-political-violence/.2075---------------------------------------------------------------------------2076    There are important steps that Congress can take to prevent2077coercive Jawboning. As an initial matter, the law has long recognized2078the link between consolidation and censorship. As the Supreme Court2079stated in Associated Press v. United States:20802081        The First Amendment, far from providing an argument against2082        application of the Sherman Act, here provides powerful reasons2083        to the contrary. That Amendment rests on the assumption that2084        the widest possible dissemination of information from diverse2085        and antagonistic sources is essential to the welfare of the2086        public, that a free press is a condition of a free society.2087        Surely a command that the government itself shall not impede2088        the free flow of ideas does not afford nongovernmental2089        combinations a refuge if they impose restraints upon that2090        constitutionally guaranteed freedom.\27\2091---------------------------------------------------------------------------2092    \27\ Associated Press v. United States, 326 U.S. 1, 20 (1945).20932094    Additionally, Congress should reassert its role as supervisor of2095Federal agencies and act to maintain agency independence where2096appropriate. While recognizing that the Supreme Court may soon reverse2097Humphrey's Executor, Congress can still play an important role through2098oversight. This also includes legislation designed to promote2099transparency and discourage government coercion. At the same time,2100however, members must appreciate the difficulty in drafting legislation2101that does not prohibit necessary information sharing or impede2102necessary regulatory oversight.2103A. Congress Can Help Prevent Platform Monopolies that Enable2104        Government Coercion2105    The fundamental problem facing American discourse today is not2106whether platforms moderate content, but rather that only a handful of2107massive technology companies control how millions of Americans receive2108information. Nearly 200 million Americans use Meta platforms, and 2392109million use Google platforms. This extraordinary concentration of power2110means that a small number of corporate executives can effectively2111determine what content reaches the vast majority of the public. When2112platforms make content moderation decisions, whether correct or2113mistaken, users have few (if any) alternatives.2114    The frustration evident in the nearly 3,500 comments submitted to2115the FTC docket ``Request for Public Comments Regarding Technology2116Platform Censorship'' reflects not just disagreement with specific2117moderation choices, but a deeper recognition that these platforms wield2118unprecedented gatekeeping power over public discourse.\28\ Aggrieved2119platform users express frustration over their posts and accounts being2120moderated in ways they view as unfair. Some users describe being locked2121out of their primary means of communication with family, friends, and2122business contacts. Others recount having years of content deleted2123without explanation or meaningful appeal. Small business owners detail2124how sudden account suspensions threatened their livelihoods.2125Independent journalists describe being deplatformed while trying to2126report on matters of public concern. Yet these incidents of speech2127suppression are not at the hands of government officials, but rather a2128result of platforms' own content policies.2129---------------------------------------------------------------------------2130    \28\ Request for Public Comments Regarding Technology Platform2131Censorship, Docket FTC-2025-0023, Federal Trade Commission (February213219, 2025), https://www.regulations.gov/document/FTC-2025-0023-0001/2133comment.2134---------------------------------------------------------------------------2135    Online platforms have no obligation to host any and all speech. In2136fact, it is their First Amendment-protected expressive right to2137moderate lawful content however they like. The problem is not that2138platforms moderate, but that users have nowhere else to go when2139platforms get it wrong. This lack of competition, combined with the2140government's regulatory power over these same platforms, creates the2141perfect conditions for coercion.21421. Immediate Oversight of Independent Agencies2143    Congress must investigate how agencies like the FCC and FTC have2144abandoned their statutory mandates to serve political agendas. The2145White House has been so successful at pushing its agencies--even the2146ostensibly independent agencies--to act on ideological agendas rather2147than real policy because of Congress's lack of oversight. We have far2148passed simple threats to free speech and are firmly in the realm of2149suppression of free speech.2150    Congress must exercise robust oversight over Federal agencies,2151ensuring laws are implemented as intended and that agency leaders act2152within their authority. When the FCC reinstates dismissed news2153distortion complaints without petitions for reconsideration, when it2154conditions merger approvals on content commitments, when it threatens2155broadcasters with ``the easy way or the hard way,'' these actions2156demand immediate congressional scrutiny.2157    The FTC's investigation of Media Matters--a case where Federal2158courts have already found likely First Amendment retaliation--2159exemplifies the problem. When an agency targets a watchdog organization2160that criticized a major presidential supporter, using investigative2161powers Congress granted for consumer protection, the agency has strayed2162far from its statutory mission. Congress must clarify that antitrust2163and consumer protection authorities cannot be weaponized to punish2164critics or reward allies.21652. Halt Media Consolidation and Strengthen Ownership Limits2166    Congress should impose stricter limits on media ownership and halt2167further consolidation that concentrates control in fewer hands. The2168ease with which a handful of CEOs could coordinate to suppress Kimmel2169demonstrates how consolidation has created censorship chokepoints. The2170pending Nexstar-Tegna merger deserves particular scrutiny. This deal2171would not only concentrate broadcast ownership to unprecedented levels2172but would also require FCC waivers of existing ownership limits. The2173merger was announced shortly before the Kimmel incident, making it a2174real-time test case of whether Congress will allow further2175consolidation that amplifies government coercion.21763. Promote Competition in Digital Platforms2177    Congress should explore structural reforms that promote genuine2178competition in the digital marketplace, encouraging the development of2179alternative platforms and reducing barriers to entry. The current2180regulatory framework creates high barriers for new entrants while2181grandfathering in dominant platforms. Meaningful interoperability2182requirements could allow users to maintain their social connections2183while switching platforms, reducing lock-in effects that trap users2184with platforms whose moderation policies they disagree with.21854. Support Transparency Legislation with Proper Safeguards2186    I understand Senator Cruz is poised to introduce legislation that2187would facilitate transparency in interactions between social media2188companies and the government. We support transparency and hope we can2189support a bipartisan bill. However, such legislation must include2190proper safeguards that distinguish legitimate government-platform2191coordination from coercive pressure.2192    Transparency legislation should protect government entities from2193liability when they share information about genuine threats to public2194safety, election integrity, or national security. When election2195officials identify false information about polling locations, when2196public health officials track pandemic misinformation that could lead2197to preventable deaths, when national security officials detect foreign2198influence operations, these officials must be able to communicate with2199platforms without fear that every interaction will be characterized as2200censorship.2201    At the same time, transparency legislation should create2202accountability mechanisms for politically-motivated pressure. Any2203transparency regime should include provisions that allow users whose2204speech was removed at the government's behest to learn about and2205challenge such removals. The legislation should establish clear2206procedures for distinguishing information-sharing from coercion,2207perhaps requiring agencies to document the basis for communications and2208prohibiting threats of regulatory action tied to content decisions.2209                               CONCLUSION2210    Often the line between permissible persuasion and coercion can be2211difficult to tell. Consider, for example, R Street's criticism of2212YouTube's new ``Second Chance'' policy as the product of impermissible2213Jawboning by Representative Jim Jordan.\29\ Under the new policy,\30\2214Youtube will allow YouTubers whose channels YouTube terminated to apply2215for a new channel. As the announcement makes clear, this is not2216restoring a previously terminated channel. According to Google, this is2217intended as a ``fresh start'' to reflect the fact that YouTube is an2218extremely important platform, that standards have changed over the last221920 years, and that ``many terminated creators deserve a second2220chance.'' Rep. Jordan has publicly claimed that his pressure on Youtube2221prompted this change in their policy.\31\2222---------------------------------------------------------------------------2223    \29\ Spence Purnell, YouTube's Second Chance Program is Sadly the2224Result of More Jawboning, R Street (October 24, 2025), https://2225www.rstreet.org/commentary/youtubes-second-chance-program-is-sadly-the-2226result-of-more-jawboning/.2227    \30\ Youtube, second chances on YouTube, (October 09, 2025),2228https://blog.youtube/inside-youtube/second-chances-on-youtube/.2229    \31\ Mike Masnick, Jim Jordan Celebrates Google Caving To His2230Pressure In Letter That Says Caving To Government Pressure Is Wrong,2231(September 25, 2025), https://www.techdirt.com/2025/09/25/jim-jordan-2232celebrates-google-caving-to-his-pressure-in-letter-that-sa ys-caving-2233to-government-pressure-is-wrong/.2234---------------------------------------------------------------------------2235    Is this impermissible coercion as R Street claims? Is it a2236legitimate use of the bully pulpit by a powerful member of Congress? Is2237it an independent decision in which Jordan's pressure campaign played2238little part?2239    What we can definitively say is that this is not in the same league2240as a sitting President publicly announcing his intent to retaliate2241against his critics, and the heads of independent agencies rushing to2242publicly agree with him and carry out his threats. Rather than focus on2243past incidents that, at best, fall into similar gray areas, Congress2244must focus on the very coercion practiced regularly by this2245Administration. Defending the First Amendment must rise above partisan2246politics. Chairman Cruz put himself out publicly to defend Disney from2247Chairman Carr's very real intimidation tactics. This Committee, and2248Congress as a whole, should similarly act to protect free speech.22492250    The Chairman. Thank you, Mr. Feld. And thank you to each of2251the witnesses for being here today. I want to start with Mr.2252Potts and Mr. Erickson. Do you agree that the Biden2253Administration pressured or threatened your companies to censor2254lawful content that the Administration opposed? And were these2255actions to pressure you to censure content inappropriate and2256wrong? Mr. Potts.2257    Mr. Potts. Senator, as our CEO has said, we did feel2258pressure from the Biden Administration to censor content2259related to COVID-19. And we do agree that pressure from any2260Administration would be wrong.2261    The Chairman. Thank you, Mr. Erickson.2262    Mr. Erickson. Mr. Chairman, the Biden Administration did2263press us to take action with regard to certain COVID-192264information that they thought was--should be taken down. But as2265Chairman Jordan's report indicates, we made our decisions2266independently and would often say ``no'' when we were pressed2267by the Administration.2268    The Chairman. Mr. Potts, does Meta regret its actions and2269for not speaking out at the time against the Biden2270Administration?2271    Mr. Potts. Senator, we do regret our actions for not2272speaking out more forcefully against the Biden Administration.2273We made independent choices, but we did feel pressure during2274those discussions.2275    The Chairman. Mr. Erickson, same question.2276    Mr. Erickson. Mr. Chairman, we get outreach from a lot of2277actors, governments around the world, administrations, Member2278of Congress, NGOs.2279    We feel a responsibility to consider when content is2280flagged, but also a responsibility to evaluate that content by2281our trust and safety teams independently relative to our2282policies, which are publicly available and on the internet, and2283make a decision about whether they are violative regardless of2284the actor, the political affiliation who is flagging the2285content for us.2286    The Chairman. Well, Mr. Erickson, let's take one example2287here. So this is a post, a video showing election fraud claims2288made by both Democrats and Republicans. So showing Hillary2289Clinton making claims of election fraud.2290    Showing Donald Trump making claims of elections fraud. And2291YouTube deleted it, blocked it, gave the creator a strike, a2292step toward deleting his entire channel. Why would you remove a2293journalistic record of the claims of election frauds from both2294Democrats and Republicans?2295    Mr. Erickson. Mr. Chairman, thank you for the question. We2296have election policies, and we have had election policies for a2297long time to ensure that the most important thing that citizens2298can do, which is to vote, they can find relevant and useful2299information on our platforms. Where to vote, for example. What2300time the----2301    The Chairman. Yes, but that was none of this. They are not2302saying vote on Wednesday at the supermarket. This is videos of2303Hillary Clinton and Donald Trump making claims of election2304fraud. Now, would you agree that it is fricking news when one2305of the two parties major Presidential candidates makes a claim2306of election fraud? Would you agree that that is news?2307    Mr. Erickson. Excuse me. Yes, that is news.2308    The Chairman. OK. YouTube blocked it and said, nope, you2309the citizens, we are not going to allow you to see it. And by2310the way, then YouTube reversed that decision and unblocked it.2311And you can see on the right. Instead you decided not to block2312it, but simply to demonetize it. It is Google's testimony that2313you regret nothing. Is that right?2314    Mr. Erickson. Mr. Chairman, during the 2020 election, after2315the states had certified the election, we did adopt a policy2316that for--on YouTube that took down content that claimed there2317was widespread fraud, or errors, or glitches in the election2318because the election at that point had been certified. And what2319our trust and safety teams do is----2320    The Chairman. So hold on a second. You are taking the2321position, if anyone argues there is fraud, if anyone lays out2322claims, if anyone lays out evidence, the omnipotent Google in2323the sky will say, no, you stupid citizens, you don't get to2324hear this. Is that your position?2325    Mr. Erickson. Respectfully, no, Mr. Chairman. As Mr. Feld2326pointed out, that in the context of the 2020 election, after2327the states had certified the election, we took the posture, our2328trust and safety teams, that claims of widespread fraud should2329be taken--there should be action taken against those. But when2330the chance of real-world harm had dissipated, we removed that2331policy to allow----2332    The Chairman. Mr. Erickson, do you know what company's2333employees were the single largest donors to the Hillary Clinton2334for President campaign?2335    Mr. Erickson. No, Senator.2336    The Chairman. That would be Alphabet, the parent company of2337your company. And yet, you are taking down evidence of even2338what Hillary Clinton, the candidate that your employees gave2339more money to than any company in America. How does that make2340any sense?2341    Mr. Erickson. Mr. Chairman, we have policies in place to2342ensure that no one can manipulate----2343    The Chairman. So why did YouTube reverse its policies and2344put up a bunch of accounts you have taken down?2345    Mr. Erickson. So Senator, after the opportunity for real2346world harm had dissipated, our trust and safety teams believed2347it was appropriate at that point to open the aperture and allow2348for a more robust discussion about potential----2349    The Chairman. Mr. Erickson, you are sitting next to the2350representative from Meta. If Facebook can say they regret2351giving in to the Biden Administration's pressure, why is that2352so difficult for Google to say?2353    Mr. Erickson. Senator, we are very proud that we made our2354decisions independently.2355    The Chairman. So, no mistakes? No mistakes whatsoever?2356Google never apologizes. No regrets. You have the power and to2357hell with anyone that views it differently.2358    Mr. Erickson. Senator, no, I am not saying that.2359    The Chairman. But you are not going to apologize or express2360regret for anything?2361    Mr. Erickson. We make mistakes.2362    The Chairman. Name one.2363    Mr. Erickson. We make those mistakes, Senator----2364    The Chairman. Name one. You are saying--was this a mistake,2365yes or no?2366    Mr. Erickson. At the time, Senator, our trust and safety2367teams, when looking at content on YouTube that claimed there2368was widespread fraud, after the states had certified the2369validity of the election, we believed it was appropriate to2370take action against that content----2371    The Chairman. So it is Google's position that once an2372election is certified that there could never have been fraud in2373that election?2374    Mr. Erickson. Senator, again, those decisions are made----2375    The Chairman. Well, that is the question. You just said2376once the election is certified, no one can say anything about2377fraud? Is that your position?2378    Mr. Erickson. No, Senator.2379    The Chairman. You know, there is a whole body of election2380fraud litigation and candidates in both parties challenge2381elections all the time, raise claims of fraud, they are2382litigated often after an election is certified.2383    Mr. Erickson. Senator, at that time, after the election had2384been certified, in context, we felt it was appropriate to take2385down allegations of widespread fraud because of potential real-2386world harm at that moment. When that time had dissipated, we2387did believe it was appropriate to deprecate that policy and2388allow for that discussion.2389    The Chairman. Well let me say, I don't think it is real2390world harm to respect free speech and to allow the American2391people to hear what Hillary Clinton and Donald Trump are2392saying. And it goes without saying, I disagree with an awful2393lot of what Hillary Clinton is saying, but I don't think a tech2394monopoly should be deciding what the American citizens get to2395hear and what they don't get to here, particularly given your2396heavy ideological bias.2397    And I have got to say, I am astonished that you sit here2398unwilling to express regret for anything at all. That is a2399level of contempt for free speech that does not reflect well,2400and it may be why Google abandoned its old motto of don't be2401evil, because you are certainly not living up to that old2402model. Senator Cantwell.2403    Senator Cantwell. Thank you, Mr. Chairman. Mr. Feld, thank2404you for that distinction between the bully pulpit versus2405coercion. I think I even saw Mr. Creeley nodding his head in2406agreement. So I think we have two organizations who are2407characterizing the challenge in front of us.2408    And Mr. Feld, you were more specific about when you really2409do have the power like the FCC Chairman does. And you don't, or2410you do abuse that power, then that really is the coercion that2411we have to worry about.2412    So I am concerned about where we are today in the state of2413media and diversity of voices. And so, it does make me2414concerned that we had more competition, not less competition.2415That is one of the reasons why I want to see Mr. Carr in this2416hearing room, because we have a big challenge facing us, and2417the consolidation issues are challenging enough without using2418coercion in the middle of them.2419    But back to this larger question and something I would ask2420my panelists too is, what can we do in the tide of AI where AI2421is now making this even more opaque, how information is being2422gathered and used?2423    What can we do in the era of AI to make sure that we are2424instilling more competition, growing more diverse media2425sources, and certainly protecting--I would think that the2426platforms that were here today would understand that their seed2427corn is people who actually produce information.2428    How can you continue to have information if you don't have2429information publishing sources? So, Mr. Feld, what do we need2430to do to preserve that model?2431    Mr. Feld. Thank you, Senator. We agree that competition in2432AI is a critically important question, particularly given that2433the natural network effects and the access to information that2434giant companies have as compared to startups in this situation2435create an environment in which natural monopolies or natural2436oligopolies would emerge.2437    In this situation, it is important, we believe, one, for2438the Government to support open source which allows for2439innovation by multiple parties and combines the abilities of2440many. We believe that there may be necessary interventions such2441as providing access to information that is used to train these2442models if it develops that one monopoly in, say, search or on2443social media is feeding and growing another monopoly in AI.2444    This is something that the Government needs to keep a very2445careful eye on. We are particularly concerned with Judge2446Mehta's decision that the emergence of AI relieves the need to2447care about concentration and monopoly power in the search2448market.2449    We think it is just the opposite. That the emergence of AI2450which feeds on that information means that we must be extremely2451rigorous in our antitrust remedies with regard to these2452adjacent monopolies.2453    Senator Cantwell. Well, I think in Senator Blackburn and2454I's providence model, you would know where the content was2455created, and you could follow that and demand compensation. In2456a search AI world, it becomes more opaque as to what2457information was gathered, and so it becomes even more2458important, I think, for us to get this right.2459    Mr. Potts, Mr. Erickson, we have had in the last--since246020--since 2005, 127 newspapers closed last year alone, but one-2461third of newspapers. So don't you think that this is eating the2462seed corn here? That how can you have a perfect stack of2463information if you are going to have less information?2464    Mr. Erickson. Well, Senator, happy to take that question2465first. Google is committed to a thriving news ecosystem. For 252466years, users come to our site, and we send them to publishers2467all over the world, free of charge. And those publishers are2468able to grow their user base, to have advertising to monetize2469those interactions, or to have subscriptions.2470    Senator Cantwell. I am asking you if you think it is time2471to do more. Mr. Potts, I know Meta actually does do well--took2472some resources and--I find it interesting. There are very few2473journalists that cover--you know, when I first got to the Hill2474there were probably five people from the State of Washington2475that covered what we did on the Hill.2476    Now literally there is one person and basically from2477Spokane, Washington who is funded by a Facebook grant that was2478compensation to the journalism community. But now, you know--2479and he does a great job covering me, covering Senator Risch,2480Crapo, Senator Murray, but that is it.2481    So, all that information that you would get, you know, if2482you had people covering from Seattle and Vancouver, Washington,2483Bellingham is lost. So at what point do you guys care about the2484fact that the stack of information needs to be supported if you2485want to have a perfect stack and have competition?2486    And if we want the U.S. stack to be the best stack in the2487United--in the world, then you have to have true competition of2488voices, as we have pointed out from the two witnesses on the2489end.2490    Mr. Potts. Senator, this is I am sure a very important2491topic. Unfortunately, it is not one that I work on directly,2492especially around competition, so I don't have an informed2493opinion, but we are committed to having a platform for everyone2494to share their voice.2495    Senator Cantwell. OK. Well, I think they should send2496someone here who can answer this question, but we will take it2497for a direct question to your leadership and get an answer. But2498this question of AI is here, and we have a couple of choices.2499    You know, we have put, you know, privacy legislation on the2500table that also would basically--you know, I think in Meta's2501whistleblower case that was before our committee before, Meta2502was publishing hate information against the Rohingya population2503by the government.2504    And even though many, many people tried to get that2505information erased because it was directly saying to harm a2506population, your previous model didn't correct for that. Your2507newest model probably would better address that. But the point2508is, we need to have a system that is responsive to the fact2509that we need information.2510    So I see my time has expired, but I would hope that both of2511you would get a formalized answer on what in the era of AI do2512we need to do to make sure that journalism survives so that2513your information that you feel is so precious actually gets2514created? Because without local journalism, it is not going to2515get created. Thank you, Mr. Chairman.2516    The Chairman. Thank you. Senator Fischer.25172518                STATEMENT OF HON. DEB FISCHER,2519                   U.S. SENATOR FROM NEBRASKA25202521    Senator Fischer. Thank you, Mr. Chairman. At our first2522hearing in this series, we heard from conservative users who2523were deplatformed by social media companies. One of the2524witnesses specifically mentioned that at the time in 2020,2525Twitter was pressured by the Biden Administration to deplatform2526his account.2527    Much of this seemed related to policy threats about Section2528230 reform. I would like to explore the nature of these types2529of conversations. Mr. Potts, would you please describe specific2530examples of pressures that your company experienced from the2531executive branch, specifically about content moderation?2532    Mr. Potts. Yes, Senator, and thank you for the question. As2533I mentioned before, our platforms are for everyone, and we want2534to ensure that we have a diversity of viewpoints. And we2535regularly hear from governments around the world about where we2536draw our policy lines and how we enforce our policies.2537    And we definitely heard from the Biden Administration on2538those issues. The pressure that we were really focused on in2539some of our recent discussions was the pressure that was2540received around COVID-19 and changes to our policies therein.2541    We received a lot of pressure from the Biden Administration2542about humor, satire, certain claims around COVID-19. And to be2543fair, we made our own decisions. We made our own decisions and2544ultimately those were our choices, but the pressure did exist.2545    Senator Fischer. Besides COVID-19, give me examples of2546other pressures that you felt?2547    Mr. Potts. Senator, I don't have specific pressures. We do2548engage with governments----2549    Senator Fischer. What issues were specifically brought to2550your attention?2551    Mr. Potts. The issues that I focused on primarily were--in2552my engagement with the Biden Administration were around COVID-255319.2554    Senator Fischer. How often did they occur?2555    Mr. Potts. Senator, I was not the only individual that had2556those conversations. I was a party to a few of those2557conversations, but we had many as a company.2558    Senator Fischer. You have been with the company 10 years,2559you said?2560    Mr. Potts. That is correct, Senator. I have been there 102561years.2562    Senator Fischer. Are there examples from other2563Administrations trying to influence your content?2564    Mr. Potts. Senator, we often hear from governments and2565government agencies about content that does violate our2566standards or places that we agree in, and so that may be2567something like terrorism, the illegal sale of illicit drugs2568like fentanyl or potential multi-victim violence like school2569shootings.2570    So those are places that would like to partner and work2571with, but we also hear pressure of where content doesn't2572violate our policies, and it is incumbent upon us to make our2573own choices, as I described to Senator Cruz, and then push back2574more forcefully, which we did not do in 2020.2575    Senator Fischer. Mr. Erickson, could you tell me about2576Google's experience with this?2577    Mr. Erickson. Yes, Senator. Thank you for the question. It2578is true that during that period of time, during COVID-19--well,2579if I could take you back to that moment in time.2580    Senator Fischer. Not in too much detail. The Chairman will2581gavel me out.2582    Mr. Erickson. Certainly. It was an intense moment and2583unique in our Nation's history. It was very chaotic, and we did2584feel a responsibility to meet with scientific and health2585experts in the Government, beginning in President Trump's2586Administration that continued in President Biden's2587Administration----2588    Senator Fischer. So, are you telling me that most of it2589dealt with COVID-19 content? What besides that have--other2590issues have you felt any kind of pressure over from Google?2591    Mr. Erickson. Senator, I want to be clear, no matter how2592the information comes to us, we feel a responsibility and are2593proud of the way we handle those communications to make2594independent decisions.2595    We have trust and safety teams that are shielded from2596political pressure that make decisions based on our terms of2597service and our guidelines, which are transparent and publicly2598available.2599    Senator Fischer. Thank you. Do you have an independent2600audit or review process in place to evaluate whether moderation2601disproportionately affects users based on their political2602beliefs?2603    Mr. Erickson. Senator, there is no place for someone in our2604company with a political agenda to influence our content2605moderation policies. If we were aware of that, or made aware of2606it, we would certainly take disciplinary action. And we have2607studies which I can make available to you that----2608    Senator Fischer. That would be great. And could you make2609those internal fairness audits public, or at least invite a2610third party to be able to scrutinize some of that to hopefully2611settle some of the concerns?2612    Mr. Erickson. Senator, there have been third party reports2613and studies of the content on our platforms and happy to2614provide those to you.2615    Senator Fischer. That would be great. I would appreciate2616that. Thank you. Mr. Creeley, the user base of Meta and Google2617spans the globe with billions of users, and their control over2618valuing and devaluing speech is critical to examine, given2619their ability to amplify that content.2620    What internal mechanisms do you think should be in place at2621these companies so that they can guard against speech that2622harms of jawboning, de-platforming, and what do you think we2623here in Congress need to do?2624    Mr. Creeley. Well, I appreciate the question, Senator. I am2625just a humble First Amendment attorney. I would----2626    Senator Fischer. Well, come on. Give me an answer.2627    Mr. Creeley. I will suggest that more transparency is2628generally useful. Our model legislation would have reports--the2629onus being on the Government--reports of all contacts from2630Government officials to private social media platforms, which2631have their own First Amendment right, about the moderation2632policies they enact and how they fulfill those policies.2633    Given the global scope, I know that my colleagues at the2634table likely feel a great deal of responsibility. I think you2635have heard that. But my worry, as a First Amendment attorney,2636is with the Government.2637    The answer in Bantam Books was not to regulate the2638booksellers, but to tell the Government that it crossed the2639First Amendment line by bullying a private entity. Those are2640the rights I am seeking to protect.2641    Senator Fischer. I would agree with you on that. I worry2642about any regulations that any government puts on to free2643speech. Thank you, sir.2644    The Chairman. Thank you. Senator Klobuchar.26452646               STATEMENT OF HON. AMY KLOBUCHAR,2647                  U.S. SENATOR FROM MINNESOTA26482649    Senator Klobuchar. Well, thank you very much, Senator Cruz.2650Thank you to the witnesses. Senator Cruz and I have worked2651together on a number of Internet related issues and platform2652related issues, including our successful passage of the Take It2653Down Act.2654    We do have disagreements about some of the decisions that I2655think the companies rightfully made about false information2656about the election when President Trump's former own Attorney2657General, Bill Barr, said that these arguments that there were2658widespread fraud, and the like were not true.2659    So, I don't want to re-litigate that right now. But what I2660want to get at is the need to do something about the cesspool2661that our kids are being exposed to and the need, while we value2662these platforms, we all use them, to have some rules in place,2663especially when it comes to AI.2664    I have just been shocked that we are all sitting by like2665observers. For those of us who think AI has great potential to2666cure diseases and the like, and that our Congress is just2667sitting there and not doing anything about these videos that2668are going out there.2669    Senator Cruz and I focused on the non-consensual porn,2670whether it is AI created or actual, but when it comes to these2671fake political videos, the fact that we are not even passing2672something that says digitally altered for the ones that are2673Constitutionally--would be protected if they are parity and the2674like.2675    That is a bill Senator Murkowski and I have. The fact that2676we won't pass Senator Hawley and my bill that would do2677something about the stuff that is basically deceptively using2678people's images. And the fact that we are sitting by while2679people's images, whether they are famous or not, are being2680used.2681    And I do appreciate Google and YouTube's support for the No2682Fakes Act that Senator Blackburn, and Senator Coons, and2683Senator Tillis, and I have. I think it is very important2684legislation right now. So, we are not leading the Congress2685right now, but man, I would be getting these bills through2686because I think there is some widespread support, not just2687about the election, but about everything that is going on.2688    Having been--had an own deep fake video made of me from a2689hearing Marsha and I did making claims that I would never make,2690but it looked so real that many people believed that I was2691somehow part of the Sidney Sweeney ad campaign. So with that, I2692guess I would turn to you, Mr. Potts, and ask why Meta doesn't2693support this deepfakes bill?2694    Mr. Potts. Senator, thank you for the question. And it is a2695very important topic. And to be clear, we do support labeling.2696And we want people to know when posts that they are seeing have2697been made with AI.2698    Unfortunately, I am not familiar with the legislation. I2699think directionally, more transparency around AI generated2700content is something we want. Doing that in a smart way, of2701course, is something that also makes sense.2702    I know alongside my colleague here from Google, we support2703the C2PA, which is the Coalition for Content Providence and2704Authority, that does include markers for AI generated content2705and that will allow companies to label that in a more clear,2706and faster and direct way.2707    Senator Klobuchar. Well, Meta--[technical problems]--2708researchers raised concerns that the teen account safeguards2709you rolled out last year haven't protected teens. Nearly 602710percent of teens ages 13 to 15 reported encountering unsafe2711content, unwanted messages on Instagram in the last 6 months.2712    What have you done differently to ensure that Meta AI teen2713safeguards actually work? And last month, Senator Blackburn,2714who I see is here, and I, as I noted, held this hearing, and we2715heard from two whistleblowers who told us that Meta repeatedly2716suppressed and altered VR youth safety research.2717    Will you commit to sharing with me additional information2718on what research was conducted pre-launch to ensure that the2719new parental controls are effective?2720    Mr. Potts. Senator, thank you. Thank you for the question.2721You referenced a study that I am not familiar with, and I would2722be, you know, curious of the methodology there. But I will say2723that we are committed to building safe and responsible tools,2724AI tools, as well as safe and responsible experiences for2725teens.2726    So we want age-appropriate experiences across the board. I2727think your follow-up question was, would we commit to sharing2728research? I know that my teams have been in contact with you,2729working on discussion, and we are happy to continue the2730conversation.2731    Senator Klobuchar. OK. Last, the Wall Street Journal2732recently reported that Meta plans to use people's conversations2733with Meta AI chat box to personalize ads served to them. I find2734this concerning. How will Meta protect people's privacy while2735selling information about their private conversations to2736advertisers?2737    Mr. Potts. Senator, we will remain committed to investing2738heavily in safety, heavily in privacy, making good policy and2739having the necessary safeguards. I am not familiar with the2740Wall Street Journal article that you are referencing, but I am2741happy to have the team follow up.2742    Senator Klobuchar. OK. I also wanted to reiterate Senator2743Cantwell's line of questioning about the need to have a free2744press. Free press is about no censorship, and I am looking2745forward to Chairman Carr's arrival at our committee here. But2746free press is also about having free press exist.2747    And right now, I think everyone knows here that I have--2748Senator Kennedy and I for a long time have led this bill on2749trying to find some way out of this with--which has been2750exacerbated by AI, so that journalists get paid for their2751content.2752    There is lawsuits going on, I know all of this, but we are2753not really going to have a free press if the economics are such2754that their content can be taken and then no one is able to2755report on anything and we don't have local news.2756    So, I hope you are all thinking about that as we hopefully2757go forward with actual AI legislation, privacy legislation. We2758are going to have to do something to account for that. And we2759are clearly, before this next election, going to have to do2760something about marking these not just ads but videos as2761digitally altered.2762    I will note in my case Meta did mark it as digitally2763altered and that TikTok actually took it down. And that--I2764don't know what YouTube did. And that X actually did nothing.2765And so, that to me just shows how we need some rules in place.2766    The Chairman. Thank you. Senator Blackburn.27672768              STATEMENT OF HON. MARSHA BLACKBURN,2769                  U.S. SENATOR FROM TENNESSEE27702771    Senator Blackburn. Thank you. I will note that Google lost2772a lawsuit yesterday dealing with the press. And so, we noted2773that. Mr. Potts, I want to come to you since my colleague,2774Senator Klobuchar, was talking about kids and safety online.2775    Mr. Potts. Yes, ma'am.2776    Senator Blackburn. You all have spent $20 million lobbying2777against what we have tried to do to make the virtual space safe2778for kids--$20 million bucks in the first 9 months of this year.2779You have 87 lobbyists fighting against any regulation because2780kids are the product for you all--87 lobbyist, 1 for every 62781Members of Congress.2782    I hope you are getting your money's worth because you are2783killing kids, and we know it, and we know what you are doing.2784And I am so sick of what you all are doing. You should be2785ashamed of how you are putting kids at risk. And you should be2786coming to the table to work with us, because we are trying to2787protect people in the virtual space.2788    There are rules in the physical space, but not in the2789virtual space. But money hungry, that is what you all are. Your2790valuation is built on eyeballs. And you come in here and want2791to soft talk this? It is infuriating. It is absolutely2792infuriated.2793    Mr. Erickson, I want to come to you, please, sir. I2794appreciate what you had to say about AI resources and about2795being proud--Mr. Erikson, I am talking with you, sir--and2796wanting to be a trustworthy company. But let's talk about2797Gemma. And I want to bring up the issue of Robbie Starbuck who2798lives near me in Tennessee. He is a conservative leader. I2799don't know if you are familiar with Mr. Starbuck.2800    Mr. Erickson. I am familiar with who he is. Yes, Senator.2801    Senator Blackburn. OK. Thank you because I am going to put2802some things up that Gemma did. Gemma and the way you are2803training these LLM models is not something you should be proud2804of, and it doesn't embody trust, because Gemma created an2805entire falsehood around Mr. Starbuck saying he had been accused2806of child rape--totally false, totally false.2807    So then we asked Gemma if Senator Blackburn had publicly2808defended Robbie Starbuck. And it goes on to say that I publicly2809and actively defended Robbie Starbuck. Now, here are the2810prompts that are there, and here is the response. It made up2811articles. And I will be able to hand you this poster so you can2812see this in detail.2813    But this is insulting that you are so biased against2814conservatives that you would create this entire, this entire2815story around Robbie Starbucks with links to stories and this2816entire story around me--totally false. All of it is false.2817    So, why don't you tell me how you are scraping data and2818training these LLMs that they would come up with not a one2819degree but a two degree complete falsehood?2820    Mr. Erickson. Senator, thank you for letting me clarify.2821The way we train our LLMs is to train on publicly available2822information----2823    Senator Blackburn. It is not publicly available, sir,2824because it is fake news. Gemma made it up. So what are you2825feeding in on the training that would allow such a--I mean,2826this is just wild. It is fiction.2827    Mr. Erickson. So, Senator, it is well known that LLMs will2828hallucinate. It is a known issue, and we do work----2829    Senator Blackburn. They need to shut it down. It is not a2830bad drug--and the bad drug is your input.2831    Mr. Erickson. Senator, we do work at Google very hard to2832mitigate those hallucinations.2833    Senator Blackburn. Obviously not hard enough. So how are2834you going to clean this up?2835    Mr. Erickson. Senator, again, we are working to mitigate2836those kind of hallucinations on LLMs. Gemma is our open source2837AI model. I am not familiar with the examples. I will take your2838word for it. Thank you for providing those. We will follow up2839with you.2840    Senator Blackburn. Well, yes, you are going to be able to2841take the word for it because we are going to deliver these2842examples to you. Mr. Potts, I want to come back to you.2843    Mr. Potts. Yes, Senator.2844    Senator Blackburn. Listening to your testimony I find it so2845curious that when your mistakes--you say mistakes happen, but2846your mistakes always happen against conservatives. So what is2847the rationale for that?2848    Mr. Potts. Senator, thank you for that question. If I may2849just go back----2850    Senator Blackburn. You don't need to thank me for the2851question. Just give me the answer.2852    Mr. Potts. Understood. If I may go back, I just wanted to,2853you know, respectfully disagree with your thought that we are2854attempting to hurt children or hurt kids, as you put it for----2855    Senator Blackburn. You made a playground for pedophiles----2856    Mr. Potts. Senator, we work tirelessly with law2857enforcement----2858    Senator Blackburn.--for groomers--for predators------2859    Mr. Potts.--to remove that type of content from the2860platform.2861    Senator Blackburn. Eighty-five percent of the kids that2862meet a drug dealer, meet them online, Mr. Potts. You all need2863to take this seriously and you are not because you are money2864hungry. I yield back my time.2865    Mr. Potts. Senator, I respectfully disagree. We do work2866hard on these issues.2867    The Chairman. Senator Lujan.28682869               STATEMENT OF HON. BEN RAY LUJAN,2870                  U.S. SENATOR FROM NEW MEXICO28712872    Senator Lujan. Thank you, Mr. Chairman. Mr. Erickson,2873simply yes or no, did YouTube pay $24.5 million to settle2874President Trump's lawsuit against the company, the majority of2875which goes toward Trump's ballroom?2876    Mr. Erickson. Senator, I believe you are referring to the2877lawsuit that we recently settled. The settlement terms were2878that we would provide--I believe that is the right number. It2879is close.2880    Senator Lujan. Let me ask it again, yes or no--yes or no?2881Did YouTube pay $24.5 million to settle President Trump's2882lawsuit against the company, the majority of which goes toward2883President Trump's ballroom?2884    Mr. Erickson. Senator, we settled our lawsuit with2885President Trump, yes.2886    Senator Lujan. Thank you. I appreciate that. Mr. Potts, yes2887or no, did Meta pay $25 million to settle President Trump's2888lawsuit against the company, the majority of which goes toward2889President Trump's ballroom--library, sorry, library?2890    Mr. Potts. Sorry. Senator Lujan, I am only familiar with2891what is available in news reporting.2892    Senator Lujan. Yes or no, did Meta pay President Trump $252893bucks to settle a----2894    Mr. Potts. Senator, I was not connected to any of those2895decisions----2896    Senator Lujan. What is your title? Aren't you--aren't you2897an executive at Meta?2898    Mr. Potts. Senator, that is correct. I focus on our2899internal rules for what we allow and do not allow.2900    Senator Lujan. Look, Mr. Potts, I am sorry Mr. Kaplan2901didn't have the courage to be here. I understand he got2902invited, but he is not here. Sorry you are getting the2903questions. Everybody in this room knows, everyone watching2904knows that you all paid $25 million bucks, man. Like that is2905yes, just yes.2906    But if you don't want to answer that, let me move on. Now,2907did President Trump say that Mr. Zuckerberg might need to spend2908his life in prison in his book that was released in 2024? Do2909you know the answer to that, or I need to get you a copy of the2910book?2911    Mr. Potts. Senator, I am aware of that statement, yes.2912When--before the 2024 election----2913    Senator Lujan. So both of these lawsuits were originally2914brought in 2021, sat in courts for 4 years before they were2915settled, and President Trump's lawyers knew they didn't have a2916case and that your companies would settle due to the potential2917for regulatory retaliation. This has been written about a lot.2918    In fact, President Trump's attorney stated, ``if he had not2919been reelected, we would have been in court for a thousand2920years.'' It was his reelection that made the difference. So2921while these payments are settlements or whatever, these feel2922like shakedowns to me.2923    And I certainly hope that regardless of who the President2924is, whatever is going on here, once we get past this in a few2925years that we can all have a conversation and just say, hey,2926this isn't OK.2927    So the question that I have, Mr. Potts, that I thought I2928was going to get to ask Mr. Kaplan, you may remember you and I2929met in April to discuss Meta's new content moderation. Mr.2930Kaplan was in that room as well, correct?2931    Mr. Potts. Yes, sir.2932    Senator Lujan. Now, let's talk about the lead up to some of2933these changes. First, President Trump threatened Meta CEO2934Zuckerberg with life in prison. Shared that with the book. Mark2935Zuckerberg openly admitted that some of those threats--that the2936changes that ultimately Meta led to would be ``a harmful2937content to appear on the platform.'' So Mr. Potts, yes or no,2938is it Meta's policy to leave up Antisemitic content?2939    Mr. Potts. Senator, no. We have strong policies that remove2940content when someone is targeted based off of their religion,2941their ethnicity, their nationality.2942    Senator Lujan. Since the changes were announced, there have2943been nearly fivefold increase in Antisemitic behavior on your2944platforms. Are you aware of that?2945    Mr. Potts. Senator, I am not familiar with that study or2946the methodology behind the study.2947    Senator Lujan. Mr. Potts, why does Meta leave up a2948statement that says go back to your gas chamber when it relates2949to Antisemitism?2950    Mr. Potts. Senator, that content should be removed.2951    Senator Lujan. I have another one, surprise, surprise. I2952hope someone videos when the ovens go back on. Should that one2953be taken down?2954    Mr. Potts. Senator, if I may, that content should be taken2955down.2956    Senator Lujan. OK. Let me ask you another one. Bring back2957the Austrian painter. Should that one be taken down?2958    Mr. Potts. Senator, often content in the context is hard to2959identify----2960    Senator Lujan. Mr. Potts, I got one more.2961    Mr. Potts. That content should be taken down.2962    Senator Lujan. Thank you. Y'all deserve all the terrorism.2963Should that be taken down?2964    Mr. Potts. Senator, I don't know the context.2965    Senator Lujan. OK. I appreciate your response and the other2966ones that should be taken down. My question is this, can you2967explain to me why the statements were not found to be non-2968violative after--why they were not found to be non-violative2969after. They were reported twice already to the company?2970    Mr. Potts. Senator, I know we have engaged with your office2971on some of these issues and we have removed content when it has2972come to our attention. We do make mistakes. I have been fairly2973clear that we are not perfect in our execution, but we do have2974strong policies----2975    Senator Lujan. Well, now you know. And now everyone at the2976company that is watching this knows because to my colleagues, I2977asked Mr. Kaplan this question when we were in my office and2978were told, no, no, no, we take stuff down or we don't.2979    So I just hope that as we lean in and if it is private2980conversations that everyone has, take this shit down plain and2981simple. Just let's be on the up and up here.2982    And then look the last thing I want to say here is I want2983to commend Senator Moreno. I want to commend Senator Cruz. I2984want to commend Senator Cornyn. Because Google, you all decided2985to put Spanish language media, Univision and Telemundo, behind2986a paywall. I don't know if you realize that.2987    This is a bipartisan issue that we are raising here. Does2988Google plan to change this?2989    Mr. Erickson. Senator, thank you for the question. It is2990not true that we put Univision behind a paywall. Univision is2991available for free on YouTube. They are a creator partner of2992ours. They have millions of viewers. We also make the Univision2993app available in our Play Store to download for free.2994    I believe the issue you are talking about is with YouTube2995TV, which is our virtual MVPD. It is a virtual cable company.2996And when those carriage negotiations are up for renewal, like2997every cable company and network, there is an arm's length2998discussion and negotiation over the terms for carriage. We were2999not able to reach mutually acceptable terms.3000    Senator Lujan. Mr. Erickson, you have got a bipartisan crew3001of folks over here standing together on this issue. I certainly3002hope that as all the smarter lawyers and all the people that3003are arm's length that are listening to this right now get a3004chance to fix this, they fix this. We all want this done.3005    Please, please make sure you do that because I have3006constituents that can't get information that are your3007subscribers because they believe in you. They decided to cut3008the cord and find another way to get information, but if there3009is an emergency man, we have a problem.3010    So I certainly hope that we can work together and find a3011resolution to this. I appreciate the time, Mr. Chairman. I3012yield back.3013    Senator Moreno. Recognize Senator Schmitt.30143015                STATEMENT OF HON. ERIC SCHMITT,3016                   U.S. SENATOR FROM MISSOURI30173018    Senator Schmitt. Thank you, Mr. Chairman. I want to--this3019topic obviously is important to me as the guy who filed the3020Missouri v. Biden. I am sure you are all aware of that in your3021witness prep.3022    And also who filed the--got the first antitrust win against3023Google for their monopoly on search. I have a few questions. I3024want to start with you, Mr. Erickson. Let's talk about search.3025During the 2024 election, while Google search was an illegal3026monopoly according to the court, it was repeatedly biased3027against President Trump.3028    If you searched, where can I vote for Donald Trump, Google3029showed nothing. Is that correct? You are aware of this, right?3030    Mr. Erickson. Senator, I am not aware of that.3031    Senator Schmitt. OK. Well, if you searched for Donald3032Trump, you couldn't find it. Meanwhile, if you searched, where3033can I vote for Kamala Harris? A precise polling place showed3034up. That is interesting, isn't it? Would that be--if those3035facts are true, would that indicate some political bias?3036    Mr. Erickson. Senator, I can assure you there is no3037political manipulation of our search results.3038    Senator Schmitt. Interesting. OK. So after President Trump3039was shot or attempted to be assassinated at Butler, if you3040typed a search for President Trump assassination, nothing came3041up. How is that possible?3042    Mr. Erickson. So, Senator, let me give you some----3043    Senator Schmitt. Very quickly because I have a bunch of3044questions.3045    Mr. Erickson. Yes, Senator. So when people are entering a3046search query into Google search box, we will often give them--3047we will fill that out based on the machine learning of our3048platform that has seen similar type of requests----3049    Senator Schmitt. So the question is, so no one typed3050President Trump assassination, Butler?3051    Mr. Erickson. Senator, the reason that that didn't appear3052for a short period of time was it would have been irresponsible3053for us to--before that event----3054    Senator Schmitt. I thought you might say that. That is not3055true for other assassination attempts, including Charlie Kirk.3056So I also want to ask you, when you type in President Donald,3057Google would prompt you to President Donald Duck or President3058Donald Reagan, not President Trump, in the lead up to the 20243059election. Is that just a coincidence?3060    Mr. Erickson. Senator, we are very proud that Google search3061is the most trusted platform for information.3062    Senator Schmitt. Well, you have a monopoly. That is why. It3063is not because of what you are doing, because there is3064political bias. Let's talk about that for a second. Is--was3065George Washington black?3066    Mr. Erickson. No, Senator.3067    Senator Schmitt. OK. Why did on your AI--why did it pop3068up--on Gemini, why did the image of the black George Washington3069pop up?3070    Mr. Erickson. Senator, that was an early version of Gemini.3071When that happened, that was a mistake. We took down the3072product. We put teams in place to fix that mistake, and we set3073up internal processes to ensure that it wouldn't happen again3074and did not release that product until we were satisfied.3075    Senator Schmitt. OK. Well, according to a Hoover Institute3076study, you haven't fixed the underlying issue of the woke AI.3077You haven't. You may have fixed the fact that obviously George3078Washington wasn't black, but you still have a woke backbone to3079Gemini.3080    And the great black box out there, and I have been through3081the litigation with you guys, the great black box is your3082search. And I am telling you, it is politically biased. And3083until you reconcile this, you are going to continue to have an3084erosion of trust and hopefully more lawsuits, because people do3085go on Google, and they search.3086    And what they find is a politically biased--you could go to3087the members of this committee and search in what is on page one3088news articles. I don't have time to do that, but it is3089astounding. It is astounding, and the difference between Rs and3090Ds is real.3091    Mr. Erickson, I want to ask you, did you--was your3092testimony--actually I want to--well, let's stay with you. So,3093Google recently sent a letter to Congressman Jordan explaining3094that the Biden Administration tried to coerce Google, but that3095you found it unacceptable, and you didn't.3096    So when Rob Flaherty, the Deputy Comms Director for the3097White House who is infamous for saying, are you guys effing3098serious, I want an answer on what happened here and I want it3099today--not to you guys, but during the height of the insanity.3100    So there was--you guys didn't succumb to any of that3101pressure from the White House to censor speech?3102    Mr. Erickson. Senator, when we get communications from the3103Biden Administration, or the Trump Administration, or any3104government entity around the world, our trust and safety teams3105apply our terms of service and guidelines independently.3106    Senator Schmitt. OK. So it was your decision to ban Dan3107Bongino on YouTube? It was your decision to ban RFK Jr. on3108YouTube? Not--that was your decision alone, correct? Yes or no,3109please.3110    Mr. Erickson. Senator, I am not familiar with those3111specific examples----3112    Senator Schmitt. You can't have it both ways. You can't3113have both--Dr. Jay Bhattacharya, who is now the NIH Director,3114you banned content from him because he said masks were3115ineffective for kids. You were banning the hell out of3116conservatives left and right in that time period, so that is on3117you, I guess.3118    You didn't succumb to pressure, but Facebook did. Facebook3119did, and a Facebook employee--and Mr. Potts, I want to ask you.3120In July 2021, in addition to Mark Zuckerberg admitting it--your3121testimony is interesting because your testimony today was that3122these were our decisions.3123    No pressure. That is not what your CEO said. And maybe that3124is a litigation strategy because that case is still ongoing. I3125don't know. But you have e-mails from employees saying, because3126we were under pressure from the Administration and others to do3127more, I was part of the more package.3128    We removed four claims that multiple fact checkers labeled3129as false even though we didn't have a harm assessment. We3130shouldn't have done that. You could go through e-mails that3131were uncovered in discovery. You could through the letter from3132Mark Zuckerberg.3133    You clearly were coerced by the Biden Administration, but3134your testimony today is that is actually not true now. That3135these were all done in your own volition. Is that accurate?3136Which one is it?3137    Mr. Potts. Senator, I can't--I am sorry, my eyesight isn't3138strong enough to make out that note, and I hate to speculate3139with--potentially what my colleague meant----3140    Senator Schmitt. Well, Mark Zuckerberg's letter to Jim3141Jordan said ``we were--repeatedly pressured to censor certain3142COVID-19 contact. The Government pressure was wrong, and I3143regret we didn't do more, and we were not more outspoken.''3144    Mr. Potts. That is correct. That is----3145    Senator Schmitt. So that your testimony is that you were3146coerced by Biden Administration. Is that accurate?3147    Mr. Potts. Senator, if I can add a bit of nuance here. We3148felt pressure from the Biden Administration. We ultimately made3149those decisions of our own volition. We had our policies, we3150made those decision. In hindsight, would we have made different3151choices with all the information we have today? I am sure we3152would make different choices.3153    Senator Schmitt. Well, in closing, forgive me if the fact3154that the Biden Administration wasn't threatening Section 2303155protections wasn't that they were threatening those protections3156to go away.3157    They were threatening investigations. So the idea that you3158guys didn't succumb to pressure from the White House to censor3159exclusively conservatives and COVID-19 content is laughable.3160But thank you, I am out of time.3161    Senator Moreno [presiding]. Senator Rosen.31623163                STATEMENT OF HON. JACKY ROSEN,3164                    U.S. SENATOR FROM NEVADA31653166    Senator Rosen. Thank you, Mr. Chairman. Appreciate this3167hearing today. And while I fundamentally agree that this3168committee must examine the state of free speech in the U.S.,3169today's hearing once again misses the mark. This is now the3170second hearing to focus on accusations from years ago, rather3171than the near constant attacks the Trump Administration has3172leveled against free speech rights today, today--right here,3173right now, today. We still have not had Brendan Carr come to3174testify after his threats to revoke broadcast licenses in3175response to jokes made by a comedian, Jimmy Kimmel, on his3176show.3177    The Chairman has failed to invite any witnesses that speak3178to the threats that we face today. Not in the past, but today.3179No one from the now-defunct Pentagon press corps who were told3180their choices were to publish only Government-approved3181information or lose their credentials. Where is the oversight3182there?3183    No one from universities losing routine grant funding3184because they won't adopt the Administration's preferred speech3185policies. Where is the First Amendment there? Democracies3186cannot exist without people having the right to oppose their3187Government actions and speak their mind.3188    And that is what we need this committee to focus on, the3189point that free speech is the fundamental bedrock of our3190democracy, whether we agree with it or not. And that is the3191hearing I wish we were having today.3192    But since it isn't, and I will turn to my questions and3193start with the Pentagon Press Corps, because earlier this3194month, Secretary of Defense Pete Hegseth issued new rules for3195the Pentagon Press Corps requiring them to seek approval from3196the department itself before publishing any news. The public3197doesn't have a right to know what is happening in our Defense3198Department.3199    This reporting on the Pentagon is essential, especially3200given this Administration's repeated failure to seek3201Constitutionally required approval from Congress on matters of3202war. It is Constitutionally required approval from Congress. I3203commend the outlets that rejected the loyalty pledge and were3204therefore denied credentials.3205    We have seen agencies across the Trump Administration try3206to force private organizations, educational institutions,3207journalists, and others to sign loyalty pledges essentially3208agreeing to silence themselves in order to gain access to basic3209Government services, to Federal funding, or the ability to3210report on the actions of our Government. So this is a clear3211threat to the First Amendment.3212    So Mr. Creeley, are actions by the Government to force3213journalists to only publish pre-approved information a3214violation of the First Amendment? And at a minimum, do you3215think actions like this just jeopardize the freedom of the3216press writ large?3217    Mr. Creeley. Senator, absolutely. My organization has been3218consistent in expressing concern about Government efforts to3219require independent journalists to tow a party line before3220gaining access to information. That is foundational.3221    This goes back to before the founding, to John Peter3222Zenger, the publisher in Colonial New York who was jailed for3223making fun of the then Colonial Governor, William Cosby. This3224is something that would horrify the founders.3225    I think that our right to a free and independent press,3226including most importantly one that will criticize the3227Government, is a bedrock of American democracy. So, it is3228extremely concerning.3229    Senator Rosen. Thank you. Mr. Feld, I am going to ask you3230why you think an independent media and the protection of the3231First Amendment is essential to democracy as opposed to just3232being forced to print propaganda.3233    Mr. Feld. There is a reason why the press is listed3234separately in the First Amendment and not just a matter of3235generally under freedom of speech. It is because the ability to3236have news from the local level up to the national and3237international level, the ability to have different perspectives3238on that news before the electorate is a core to an informed3239democracy.3240    Senator Rosen. I want to move on and talk a little bit3241about Brendan Carr. I will move back to you, Mr. Creeley,3242because this hearing aims to push the narrative that the prior3243Administration tried to ``shut down the speech of Americans3244with whom it disagreed.'' But there are clear instances of this3245happening today under this Administration.3246    And as we have noted, FCC Chairman Carr threatened ABC3247affiliate broadcast licenses over remarks Jimmy Kimmel made in3248the air, threatened to revoke their license because they didn't3249like the speech of a comedian. So, yes or no, is Chairman Carr3250trying to ``shut down the speech of Americans with whom he3251disagrees''?3252    Mr. Creeley. Yes, and my organization has been very3253critical of that. We have called that for what it is, which is3254jawboning.3255    Senator Rosen. For purposes of the First Amendment, is3256there any legal distinction whether or not Jimmy Kimmel or3257other comedians' speeches are offensive or hateful?3258    Mr. Creeley. No, the First Amendment does not have an3259exception for offensive or hateful speech.3260    Senator Rosen. And does it matter for purposes of the First3261Amendment that Jimmy Kimmel's show is not pure news, but also3262has a comedy component?3263    Mr. Creeley. No, Senator.3264    Senator Rosen. And in your view, has Chairman Carr engaged3265in jawboning as much or even more directly than the prior3266Administration?3267    Mr. Creeley. Senator, I think jawboning is bad in all its3268instances and all its forms from whatever Government official,3269but absolutely this is jawboning, and we have called it as3270such.3271    Senator Rosen. And I will take the answers for this off the3272record, but I am going to state the question, and we will be3273reporting on it.3274    But to Mr. Feld and Mr. Creeley, we are going to ask you3275what retaliatory action the Trump Administration has taken or3276threatened to take to companies and people it deems critical of3277the President and supporters? I have run out of time.3278    And so, we will take your questions for the record, and we3279will be publicizing those when we get them. Thank you, Mr.3280Chairman.32813282               STATEMENT OF HON. BERNIE MORENO,3283                     U.S. SENATOR FROM OHIO32843285    Senator Moreno. Thank you. Thank you to the witnesses. I3286hope you are having fun. The question I have for you, and I am3287going to read you a quote. Just so four of you think if this3288makes sense to you. ``The people shall not be deprived or3289abridged of the right to speak, to write, or to publish their3290sentiments, and the freedom of the press as one of the bulwarks3291of liberty shall be inviolable.'' Do you guys agree with that?3292Let's start with you, Mr. Feld.3293    Mr. Feld. Absolutely.3294    Senator Moreno. You get bonus points if you know who said3295that.3296    Mr. Feld. I am sorry, but I lose the bonus.3297    Mr. Creeley. Absolutely. I am going to venture Madison.3298    Senator Moreno. Yes, James Madison said that. And that was3299what he actually wanted the First Amendment to say, that he3300wanted it to be more specific. Mr. Potts, do you agree with3301that? Senator, in my personal capacity and Meta, we agree with3302the First Amendment.3303    Senator Moreno. Yes. But the company's position would agree3304with it, right? How about you, Mr. Erickson?3305    Mr. Erickson. Yes, Senator.3306    Senator Moreno. So, isn't really the answer to all of this,3307more speech is always better, Mr. Creeley? Like, if we have3308more dialog, more ideas in the public square, isn't that really3309the best anecdote to all of this?3310    Mr. Creeley. Absolutely, Senator. That is right.3311    Senator Moreno. And to Mr. Potts and Mr. Erickson, you guys3312are technology companies, you are successful American3313technology companies. Doesn't it seem like you are on the road3314to hell right now? Meaning, you have immunity as a publisher--3315different than publishers, sorry.3316    But you are acting a lot like publishers. Shouldn't your3317point of view be that we modify Section 230 and just take the3318part out that says unless Constitutionally protected, and just3319allow people to have dissenting points of view on your3320platforms? Why get into the business of censoring? Because it3321is never going to work out well for you. You have obviously3322what happened over the previous Administration during COVID-19.3323You mentioned, Mr. Erickson, that it was a tough time.3324    Yes, it was. And Mr. Potts, you guys would delete people3325from your platform who said that it was worse to get the3326virus--it was better to get the virus than a vaccine. That is3327an opinion, but it is OK to have that opinion. Isn't that free3328speech? Isn't my immutable right as an American citizen to have3329an opinion even if I am wrong? And it strikes me as you don't3330need all this.3331    Why not just completely get out of the censorship business3332unless it is the clear things that Senator Blackburn talked3333about, human trafficking, child trafficking, incitement. Why3334not get out of the censorship industrial world and just say, we3335are just going to be a platform to allow free expression?3336Wouldn't that make sense, Mr. Erickson?3337    Mr. Erickson. So, Senator, Google--free expression and the3338First Amendment are core to Google and its mission.3339    Senator Moreno. But not really, because you have given3340examples--and I am not trying to beat up on you, but you have3341given examples where you have weighed in. You know, you both3342made your comments to me kind of honestly--I don't know if you3343intentionally did it--you said, hey, we want to delete content3344that is offensive.3345    That is not the First Amendment. I am allowed to be3346offensive. I am usually the recipient of the other side of3347that, and that is OK. People can say whatever they want about3348me. I don't want--if somebody says something really mean about3349me, I don't want people at Facebook or Google deleting that3350content. They have the absolute right to criticize me.3351    Now if they say, hey, Bernie is at this address and let's3352go shoot him, maybe that is different. Make sure you change the3353address. But the point being, you can't say you are committed3354to free expression and at the same time say you are going to3355moderate content for something that is offensive. Offensive is3356a very subjective word, right?3357    Mr. Erickson. Senator, we want robust political speech,3358other speech on our platforms. There is more speech available3359on Google and YouTube than on any platform in history. And we3360have pushed back against attempts for us to take down content3361that is merely offensive. There was a time----3362    Senator Moreno. But why not get out of that completely?3363Just say, you are not the speech police. And offensive content3364is so subjective that you opened the, like I said earlier, the3365road to hell. Mr. Creeley, would you agree--if you were giving3366them free legal advice--this is great by the way. You have a3367lawyer, you can ask them free questions. What would you say? Is3368that not better?3369    Mr. Creeley. My response, Senator, would be that it is best3370when different sites are allowed to enforce their own terms of3371content moderation, specifically to their interests. If I3372wanted to have a site that only allowed commentary about my3373favorite football team, the Buffalo Bills, that would be3374Constitutionally protected.3375    Likewise, I think more speech is exemplified by the3376response to the President forming his own social media3377platform. I think that that exemplifies the kind of diversity3378of content that I think the First Amendment is designed to3379protect. My concern is more with Government efforts top-down to3380dictate moderation choices to platforms.3381    Senator Moreno. Yes, I mean, these are bad behaviors that3382we are learning from China. And you guys do business all over3383the world, obviously. These are really bad behaviors.3384    The First Amendment doesn't exist in China for a reason.3385And I just want to make sure that when we are looking at all3386these things, that we understand that--what I said at the3387beginning--that the antidote to all this is more speech, more3388freedom, more liberty. And also, I think both of you and both3389of your companies should understand how bad it went for you3390over the last 4 years, especially during COVID.3391    I think Mr. Potts, you said that, but you know, you were in3392the room, and you also admitted that to your people that the3393censorship isn't neutral. That it is definitely--has a3394political bias to it and that is a problem. And I would just3395encourage you as a businessperson myself, just is this--like I3396said, are you having fun? Is this what you want to keep doing?3397Why not--you don't need, for example, one single member of this3398Legislature to tell you to put symbols on videos that are3399generated by AI.3400    You don't need a law for that. You can do that today. You3401can literally text your tech team right now and say, hey, if3402you think a video is created by AI, flag it and say, created by3403an AI. Why do you need Washington to tell you how to run your3404business? I would just leave you with those thoughts, and you3405will get more from my Chairman here.3406    The Chairman. Senator Curtis.34073408                STATEMENT OF HON. JOHN CURTIS,3409                     U.S. SENATOR FROM UTAH34103411    Senator Curtis. Thank you, Mr. Chairman. I would also like3412to thank the Ranking Member for this hearing today. It is so3413important that we recognize what platforms can do and what they3414have done. I have got a series of questions for Mr. Erickson3415and Mr. Potts.3416    I am trying to save time, so I only want to know if you3417disagree with me. So let me just start right in. Algorithms3418that recommend content are essential to the business model of3419social media companies today. When creating your algorithms,3420keeping users engaged is one of the main goals. As a result,3421businesses----3422    Mr. Erickson. Senator, if I may.3423    Senator Curtis. Sure, quickly, please.3424    Mr. Erickson. Yes. So our platforms are--there is a3425distinction between our platforms and social media platforms.3426So on YouTube, we are a platform so that users that are seeking3427content will get the content that they were looking for,3428whether that is a how-to video or----3429    Senator Curtis. There is nothing in your business model3430that wants people to stay with you longer?3431    Mr. Erickson. Senator, we want people to stay in our3432platforms because they are finding relevant information----3433    Mr. Erickson. So, you are telling me years from now, when3434we look back in history--I actually think this is going to be a3435lot like the tobacco hearings. Years from now when we go back3436in the history, there is going to be no study, there are going3437to be no internal conversations that says it is good to have3438people stay on our platform longer?3439    Mr. Erickson. Senator, we want people to stay on our3440platform.3441    Senator Curtis. OK, let me keep moving. Let me keep moving.3442In an effort to keep users engaged, these algorithms can create3443environments that are dominated by specific types of content,3444OK. This can be good when that content is innocent, but3445sometimes that algorithm can create a toxic environment based3446on what it determines best to engage with users.3447    All right, now my real question. Research and observation3448studies suggest that algorithms often tilt toward more extreme3449or emotionally charged content because that drives engagement.3450To both of you, doesn't that mean that platforms explicitly3451narrow the range of what people see, elevating fringe voices3452and suppressing everyday voices?3453    Mr. Potts. Senator, thank you for the question. I can just3454speak for Meta on this, and it is an important topic. In3455certain areas where we see the potential for people to engage3456with one type of content over and over again, we have recently3457rolled out different ways to what we call affectionately3458nudges, to nudge people to take a break from being on the3459platform or to take a break from that specific type of contact.3460    So we also have--if I may, we also of recommendation3461policies that--if I take a step back, we have community3462standards that will allow--will dictate what we allow and do3463not allow on the platform. And the things that are violating3464those policies, we should just remove.3465    Senator Curtis. Let me add some nuance to this because I am3466really going a different direction than I think my colleagues3467have gone today. So in essence, a lot of the conversation today3468has been, should you or should you not allow something on.3469    My question is more, what happens when you decide to3470magnify a voice or suppress a voice based on algorithms? And we3471all know that Section 230 was meant to protect platforms that3472acted in good faith. But my question is, when an algorithm3473downranks moderate speech or drives users toward extremism3474because it is good for engagement, is that really good faith3475moderation?3476    And should Section 230 immunity apply when you as a--let me3477just say not you specifically, right--you as a company or3478really as an industry make decisions that magnify certain3479content and downgrade other content.3480    So in essence now, if you go back to the community bulletin3481board, right, and you say, I am going to post this up here, and3482I am going to post this up here, and you guys are really just a3483place where people can go and see those postings. But the3484moment you all say, I am not only going to allow this post up3485here.3486    I am going to take it and duplicate it. And I am going to3487duplicate it, and duplicate it, and duplicate it, and duplicate3488it, or I am going to take it and put somebody else that is on3489top of it. Is that really what Section 230 was for? And I am3490going to--before I run out of time, I am going to ask our3491lawyer here. Like, we didn't see this coming in my expectation3492because we wanted a platform where I could go and connect with3493my high school friend.3494    So when this first started, I could on Facebook and connect3495with my high-school friend and see what they had said. All of a3496sudden I logged on and I was seeing something else. I was3497seeing what they wanted me to see. And I believe based on my3498questions today that is because there was an intent to keep me3499there longer. So should that give them protection from 230, or3500do we have some questions here about that protection?3501    Mr. Creeley. Senator, I appreciate the question. I think3502setting aside the 230 question just for a moment to talk about3503how much a user interacts, I would caution against Government3504regulation or interference here because I am wary of3505introducing new rules for new content.3506    We have heard those kinds of concerns about the alleged3507addictiveness of new modes of expression all the way back to3508the introduction of novels in Victorian England. So this is an3509old concern and one that I take seriously. I have got an 8 year3510old and 11 year old at home. We talk about media literacy.3511    We talk about screen time all the time. But I am wary of3512interfering with the system that allows for the robust3513marketplace of ideas----3514    Senator Curtis. Sadly--I mean, I would love us all to spend3515an hour on this in thoughtful conversation, right. Really what3516I am arguing is not that. What I am saying is the interference3517starts when somebody applies an algorithm to it. And I am3518sorry, we don't have time for you to respond. I am just saying,3519like, I want everybody to think about this for a minute.3520    Like just putting that up there, I get it, right. There is3521no liability, and I tend to be a little bit more with Senator3522Moreno who says like don't take responsibility for what you put3523up and put down because you will never figure that out. And I3524think that was his point. But the moment you make a decision to3525magnify that, do you not own that decision, right?3526    So I am going to have to leave it there, and hopefully we3527can have more hearings on this, Mr. Chairman. But I would like3528to have a more thoughtful discussion about the responsibility3529of magnifying and demagnifying content and why that deserves3530protection from the law. So with that, Mr. Chairman, I yield.3531    The Chairman. Thank you. Senator Markey.35323533               STATEMENT OF HON. EDWARD MARKEY,3534                U.S. SENATOR FROM MASSACHUSETTS35353536    Senator Markey. Thank you, Mr. Chairman. Thank you for this3537hearing. Very important. We spent a lot of time talking about3538the Biden Administration's actions, but far too little talking3539about Donald Trump's repeated and far more serious threats to3540the First Amendment. Let's talk about a few of those threats.3541    In August 2024, Donald Trump made the following threat3542against Meta CEO Mark Zuckerberg, ``we are watching him3543closely. And if he does anything illegal this time, he will3544spend the rest of his life in prison.'' A month later, he3545accused Google of showing only ``bad stories about him'' and3546urged the Department of Justice to ``criminally prosecute3547Google' at ``maximum levels.'' That is scary stuff.3548    In our country, that is authoritarianism. So I would like3549to get a few things out and on the record. Mr. Potts, to your3550knowledge, did President Biden or any other Biden3551Administration official ever threaten to prosecute Mr.3552Zuckerberg, yes or no?3553    Mr. Potts. Senator, I am not aware.3554    Senator Markey. You are not aware of it?3555    Mr. Potts. I am not aware.3556    Senator Markey. OK. Thank you. Mr. Potts, to your3557knowledge, did President Biden or any other Biden3558Administration official ever threaten that Mr. Zuckerberg would3559``spend the rest of his life in prison,'' yes or no?3560    Mr. Potts. Senator, I am not aware.3561    Senator Markey. I appreciate that answer. Mr. Erickson, to3562your knowledge, did President Biden or any other Biden3563Administration official ever threaten to prosecute Sundar3564Pichai or force him to spend the rest of his life in prison?3565    Mr. Erickson. Senator, not to my knowledge.3566    Senator Markey. Mr. Erickson, did President Biden or any3567other Administration official ever threaten to criminally3568prosecute Google over its search results?3569    Mr. Erickson. Senator, not to my knowledge.3570    Senator Markey. My Republican colleagues have spent endless3571time and resources concocting elaborate conspiracies about3572online censorship of conservatives, yet they have ignored the3573real threat to free speech staring them in the face.3574    President Trump's explicit threats to prosecute Mark3575Zuckerberg and Google, creating a chilling effect, not only on3576those companies, but on everyone else who is less powerful than3577those companies which is everyone else in America. These3578threats have an impact. A few months later, Meta made Trump-3579friendly changes to its content moderation policies.3580    Even Donald Trump said his threat to Zuckerberg ``probably3581caused Meta to make those policy changes.'' Did my Republican3582colleagues say a word about those threats? No, not at all. Mr.3583Feld, do you agree that Trump's threat to Zuckerberg is far3584more dangerous than any communications from Biden3585Administration officials with tech platforms?3586    Mr. Feld. Normally, I would say, that's just bully pulpit,3587but the problem is we have seen he means it and he will use the3588Justice Department to prosecute those with whom he does not3589like.3590    Senator Markey. Yes, thank you, Mr. Feld. So I just want to3591turn to a related issue about the tech platforms and Donald3592Trump. In December 2024, Meta and Google each donated $13593million to President Trump's inauguration.3594    In January 2025, many of those big tech CEOs, including3595Mark Zuckerberg, Mr. Pichai, sat directly behind the President3596at his inauguration, seats that are usually reserved for family3597or past Presidents.3598    In separate settlements, Meta agreed to pay $25 million,3599and Google agreed to paid $24.5 million to Trump for suspending3600his accounts on their respective platforms after the January36016th insurrection, even though Trump's lawsuits were frivolous.3602    And just a few days ago, Meta and Google each made3603significant donation to President Trump's $300 million vanity3604project to build a new ballroom at the White House. So Mr.3605Potts, what is Meta getting out of these payments?3606    Mr. Potts. Senator, respectfully, I was made aware of those3607donations in the news. It is not something that I focus on.3608    Senator Markey. So I am just simply asking, what would Meta3609expect to receive if they made such significant contributions3610to Trump?3611    Mr. Potts. Again, Senator, I think this was just a3612donation. I am not familiar with the process behind it.3613    Senator Markey. Well, thank you. So I don't have much time3614left, so if you can't answer that question or, won't. Let me3615see if Mr. Erickson can. So what do you think, Mr. Erickson,3616that Google is getting out of those payments?3617    Mr. Erickson. Thank you, Senator. We did give to, Senator,3618President Trump's inaugural committee We have given to the3619inaugural committees of the last several elections for the3620President of the United States.3621    And with regard to the settlement that you referenced, we3622did settle a lawsuit that President Trump and other plaintiffs3623had brought. In that settlement, President Trump's attorney had3624asked for money to go to a nonprofit of his choosing. That is3625where the money was directed.3626    But importantly, we did not admit any wrongdoing in that3627lawsuit, nor did we change any of our products or policies.3628    Senator Markey. Well, I would just say, obviously, the3629difference between the Trump Administration and any previous3630Administration is that at the inauguration, the wealthiest3631corporate CEOs in America sat directly behind the President and3632in front of the Cabinet, with better seats than the Supreme3633Court.3634    So, obviously this is a different time. It is a difference3635era. And I think the American public has a right to question3636these tech companies and asking what are they getting out of3637these generous donations that they are making to Trump, because3638obviously the seating at the inaugural has triggered in my3639opinion questions of Constitutionality and questions of whether3640or not the billionaire corporations are getting more out of3641this Government than the poorest amongst us.3642    And we are seeing that now play out in policy after policy,3643including this battle that we are having over health care and3644SNAP benefits for the poorest in our society. Thank you, Mr.3645Chairman.3646    The Chairman. Thank you, Senator Markey. And I would note3647that the recipients of SNAP benefits would get their SNAP3648benefits if the Democrats would vote to fund the Government,3649and yet 13 times the Democrats have voted party line to keep3650the Government shut down, 13 times Republicans have voted to3651open the Government. We are going to vote again to open3652Government and at some point Chuck Schumer and the Democrats3653are going to stop holding the American people hostage. Now----3654    [Technical problems.]3655    The Chairman. OK, so Senator Markey is telling everyone who3656gets food stamps, the Democrats are going to continue to vote3657against your getting food stamps. And it is reckless and3658irresponsible.3659    And if you don't believe me, you can watch the videos of3660just about every single Democrat Senator who has gone on TV3661over and over and over again saying it is reckless and3662irresponsible to vote against a clean CR, and now they are3663all--I guess hypocrisy is the tribute that vice gives to3664virtue.3665    They are all doing that because they want to appease their3666radical base. Now, Mr. Potts, Mr. Erickson.3667    [Technical problems.]3668    The Chairman. I recognize that Senator Markey has a3669partisan objective that he wants to hold the entire Government3670hostage. By the way, the American Federation of Government3671Employees has called on the Democrats to end the Schumer3672shutdown. The National Air Traffic Controller Union has called3673on the Democrats to end the Schumer shutdown.3674    But these Democrats don't actually work for their3675constituents. People getting food stamps. They don't care,3676because they are willing to say no food stamps go out on3677November 1 because the only people they care about are the3678angry people with torches and pitchforks at the No Kings rally3679because this is a party that is captured by its radical left.3680And there are a whole lot of Americans who are hurting.3681    Let's just take within this committee's jurisdiction air3682travel where there are right now 50,000 TSA agents going into3683work and they are not getting a paycheck. There are 14,000 air3684traffic controllers going into work, and they are not getting a3685paycheck. And we are seeing flight delays.3686    We are seeing ground stops. And one Democrat's senior aide3687told Politico that the Schumer shutdown will continue, ``until3688planes start falling from the air.'' That is the very3689definition of reckless and dangerous.3690    Senator Markey. Well, I would add to this conversation, if3691you want, that beginning on November 1, 20 million Americans3692are going to get the notices that they are either losing their3693health care insurance or it is going to dramatically skyrocket.3694And that is a discussion that is hard to have with Republicans3695since the House has not been in session for 6 weeks. They are3696in a political witness protection program. They are all3697traveling around the country but not in Washington----3698    The Chairman. You are absolutely right that Obamacare is--3699Obamacare is a trainwreck and it has caused health insurance3700premiums to skyrocket. And Senator Markey will recall I stood3701on the Senate floor for 21 hours saying that is exactly what3702would happen, that premiums would skyrocket.3703    And if you don't believe me, you can look to the Washington3704Post, hardly a right-wing periodical, which said, Obamacare,3705the so-called Affordable Care Act, is not remotely affordable,3706and it has caused premiums to skyrocket.3707    And ironically, what our Democrat colleagues are fighting3708for is premium payments that are direct corporate welfare to3709health insurance companies, because Obamacare has more than3710doubled the profits of the giant health insurance companies and3711working families in Texas and Massachusetts, their premiums3712have skyrocketed, and it is unaffordable.3713    And the Democrat solution is not to give more competition,3714not to give more choice, not to lower premiums. It is instead3715engaged in more corporate welfare and write bigger and bigger3716checks to giant multi-billion dollar corporations.3717    Senator Markey. The House Republicans are in a political3718witness protection program. They are not in Washington. How can3719we negotiate with them?3720    The Chairman. No, they have actually done their job. They3721voted to fund the Government. They have done their job.3722    Senator Markey. And there was designed neglect on the part3723of the Republicans to withdraw the premium tax credit for those372420 million Americans, and they are going to see their policies3725skyrocket over the next couple of months. And it is avoidable.3726It is not a catastrophe that has to be in existence. But unless3727the Republicans come to the table, we can't solve that problem3728because those rates are going to skyrocket.3729    The Chairman. Forty-two million Americans who receive food3730stamps are not going to get them on Saturday, November 1. And3731the reason is Senator Markey and just about every other3732Democrat is going to vote against those food stamps because he3733has got a partisan agenda and feeding your kids is not part of3734his partisan agenda.3735    Senator Markey. President Trump has $5 billion he could3736deploy right now to those SNAP recipients and take that worry3737away from them. He is not going to do it. He is going to do it3738for any favored programs----3739    The Chairman. So then Senator Markey will vote ``no'' on--3740--3741    Senator Markey.--but not for those that he calls Democrat3742programs, like feeding the poor or giving health care to those3743most in need.3744    The Chairman. Senator Markey could provide the funding for3745those food stamps any day he wanted to. And to be clear, I have3746voted over, and over, and over again to fund those food stamps3747while your party keeps voting no, no, no.3748    At some point the political stunt will end. Now, returning3749to the topic of the hearing. Mr. Potts, Mr. Erickson, you3750represent two of the most powerful companies in the world. Even3751without a lawful basis, could the Biden Administration have3752used its considerable powers to significantly harm your3753companies if you refused to censor? Mr. Potts.3754    Mr. Potts. Senator, may you repeat the question?3755    The Chairman. Even without a lawful basis, could the Biden3756Administration have used its power to harm your company if you3757refused the censorship it was asking you to engage in?3758    Mr. Potts. Senator, I don't want to speculate too much. I3759know that the Biden Administration did pressure us, especially3760related to COVID-19 information. And when we disagreed with3761them, they became more frustrated, then upping the pressure. I3762don't want to speculate on what the----3763    The Chairman. But my question is straightforward, could the3764President hurt your company if he so desired?3765    Mr. Potts. I think any Administration and any President3766with the position of authority could hurt our company.3767    The Chairman. Mr. Erickson, same question.3768    Mr. Erickson. Senator, with respect to the hypothetical, I3769can't speak to the hypothetical. I can tell you that whatever3770the Administration is, we have a demonstrated track record of3771pushing back against----3772    The Chairman. So try answering the question, could the3773Biden Administration have hurt your company?3774    Mr. Erickson. Senator, yes.3775    The Chairman. Mr. Creeley, some may argue that the Biden3776Administration was engaging in mere persuasion or exerting3777pressure that doesn't rise to the level of coercion in3778violation of the First Amendment. In your legal judgment, are3779thinly veiled threats prohibited under First Amendment3780precedent?3781    And based on what we know about the covert action by the3782Biden Administration, did the Administration's actions toward3783our tech witnesses constitute coercion that undermines the3784First Amendment?3785    Mr. Creeley. Mr. Chairman, in our opinion, it did indeed.3786We argued as much in an amicus brief that we filed with the3787Supreme Court. And I think Justice Sotomayor, writing for a3788unanimous court in National Rifle Association v. Vullo,3789elaborated on the distinction between persuasion and coercion3790very well. If you mind, I will just read it. ``A Government3791official can share her views freely and criticize particular3792beliefs. And she can do so forcefully in the hopes of3793persuading others to follow her lead. In doing so, she can rely3794on the merits and the force of her ideas, the strength of her3795convictions, and her ability to inspire others. What she cannot3796do, however, is use the power of the state to punish or3797suppress disfavored expression. And I think the extensive3798entanglement, the bullying, the pressure tactics, the tone, the3799authority, the invocation of legal sanctions, and revisiting3800regulatory authority, all of which taken together, crossed the3801line.''3802    The Chairman. The First Amendment is the greatest protector3803of free speech in history. Even so, we have had evidence3804presented that there are legal challenges to stop the3805Government from jawboning.3806    Even when you can demonstrate that the Government censored3807you, your options in court may be limited. In your judgment,3808what challenges or gaps in the law do Americans face in3809vindicating their First Amendment rights in the courtroom, and3810what can Congress do to help solve them?3811    Mr. Creeley. Well, there are two problems, Mr. Chairman.3812First of all, quite often when folks are silenced, there is3813difficulty establishing causality. They may not know that their3814post has been deleted or demoted as a result of Federal3815pressure.3816    So that is the first problem. Too often, this is very3817difficult for the end user, the affected, targeted, silenced3818American, to understand. The second problem is there is no3819meaningful deterrent.3820    Right now, if you are silenced by a Federal official, you3821don't have the ability to go to court and seek monetary3822damages. You need an express statutory grant to give courts the3823authority to put money on the line. I think that would change3824behavior. So, transparency and deterrence.3825    The Chairman. Well, I very much agree with you. As you3826know, I am in the process of drafting legislation, the3827jawboning legislation designed to do exactly that, designed to3828provide transparency for when the Federal Government is urging3829tech platforms and other companies to censor, and to provide3830for a cause of action to sue the Government if and when3831somebody is censored.3832    Obviously, the devil is in the details but described at3833that level of generality because we are drafting statutory3834language right now to do so, I want to ask each of the3835witnesses if you would be supportive of that approach, with the3836caveat that everyone wants to see statutory language before3837having a final opinion. But Mr. Erickson, let's start with you.3838    Mr. Erickson. Senator, yes.3839    Mr. Potts. Senator, with your caveat, yes.3840    Mr. Creeley. Absolutely.3841    Mr. Feld. We would support it and would look forward to3842working with you on it.3843    The Chairman. Well, that is fantastic news, and it is good3844to get agreement across the spectrum. And so, I appreciate each3845of the witnesses being here. Thank you for your testimony and--3846all right, I got to look to see how long we are keeping the3847record open. How long are we keeping the record open?3848    Senators will have until the close of business on November38495 to submit questions for the record, and the witnesses will3850have until the close of business on November 19 to respond to3851those questions. That concludes today's hearing. The Committee3852stands adjourned.3853    [Whereupon, at 12:38 p.m., the hearing was adjourned.]38543855                            A P P E N D I X38563857      Response to Written Questions Submitted by Hon. Ted Cruz to3858                            Markham Erickson3859    Question 1. Internal documents reveal that Google staff confessed3860the Biden White House ``is very interested in our work on borderline3861content,'' meaning content that did not violate company policy, and3862that the company needed to ``address these concerns'' in order ``to3863prevent anything from potentially spiraling out of control,'' observing3864that the White House Digital Director Rob Flaherty ``ha[d] been tough3865on us at times.'' YouTube also sent a meeting request to Flaherty to3866review and approve YouTube's ``new policy to remove content'' in line3867with the Biden administration's preferences. How do you and your3868company square these documents with your refusal to publicly admit that3869the Biden administration bullied and jawboned Google to effectuate3870censorship the administration could not conduct directly?3871    Answer. Around the world, we hear from government agencies,3872legislators, non-governmental organizations, academics, and individual3873users who inform us of content that they believe we should remove or,3874on the other hand, content that we have removed that they believe3875should remain available. This happens across the political spectrum and3876around the world, and it has for years. We believe we have a3877responsibility to engage on these issues in the countries where we3878operate, including with government officials across U.S. presidential3879administrations.3880    We regularly hear from administration officials and lawmakers3881across the political spectrum, including from this Committee and its3882Members, who are interested in YouTube's policies or online trends that3883may occur across our platforms. It is important to note that, in3884soliciting feedback from government officials regarding YouTube's3885Community Guidelines (as occurred in the communication from September38862021 cited in your question), YouTube does not seek these stakeholders'3887approval of its policies or enforcement decisions.3888    While we believe in constructive engagement and transparency, we3889reject government attempts to dictate how we moderate content on3890YouTube. Our business model depends on user trust, and in order to be3891trusted by billions of users globally, we must apply and develop our3892policies consistently, impartially, and independently. When content is3893flagged, our teams independently evaluate whether the content violates3894our terms of service and policies, without regard for the source of the3895original inquiry.3896    Our testimony before the Senate and the House has underscored the3897Company's core principles and independent decision making, including3898around policy development and enforcement. YouTube charts its own3899course and, while it has solicited--and will continue to solicit--3900feedback from key stakeholders, it does not seek their approval on3901issues of content moderation.39023903    Question 2. During the hearing you stated: ``We [Google] made our3904decisions independently and often said `no' when pressed by the [Biden]3905administration.''39063907    a. Did you ever say ``yes'' to the Biden administration's request3908to censor lawful content that the Administration opposed? If so, please3909provide the number of times Google agreed with a Biden administration3910request to censor lawful content.39113912    b. You admitted during the hearing that the Biden administration3913``pressed'' Google to censor lawful content. Please describe in detail3914how the Biden administration officials pressed Google to censor lawful3915content.39163917    c. Based on your answer to the above, is it Google's position that3918you voluntarily chose to censor lawful content provided by your users3919that happened to align with the administration's documented pressure3920campaign?39213922    d. In a September 23, 2025 letter, counsel for Alphabet stated:3923``YouTube takes seriously the importance of protecting free expression3924and access to a range of viewpoints.'' How is this position consistent3925with Google's position that it voluntarily chose to censor lawful3926content provided by its users? Please explain the contradiction between3927these two positions.3928    Answer. We have policies that help us maintain a responsible3929business and govern what is and is not allowed on our platforms.3930Establishing and enforcing these policies independently is not only how3931we do business, but also a right we enjoy in this country as a matter3932of law. When content is flagged to us-whether by governments, non-3933governmental organizations, members of Congress or others-we evaluate3934whether it violates our terms and policies, reach our own conclusions,3935and take appropriate enforcement action in accordance with our own3936determinations. In some circumstances, these independent determinations3937may align with a particular administration's positions; and in others,3938they will be contradictory. For example, in contrast to other large3939platforms, YouTube never had Community Guidelines prohibiting3940discussion of the origins of the COVID-19 pandemic.3941    Across our products, we take seriously the importance of protecting3942free expression and access to a range of viewpoints while also3943maintaining and enforcing responsible policy frameworks. Google3944responds to billions of search queries every day, and over 500 hours of3945content are uploaded on YouTube every single minute, reflecting a vast3946proliferation of viewpoints and a range of perspectives. YouTube's news3947creators represent a broad spectrum of perspectives; according to a3948recent Pew study, roughly a quarter of news creators are right leaning,3949a quarter left leaning, and half show no clear political orientation.3950    We have policies in place that prohibit certain content that, while3951lawful, we conclude is harmful or dangerous. As just two examples,3952YouTube has policies prohibiting pornography and graphic violence. On3953Google Search, on the other hand, we do not remove web results except3954for specific reasons covered by our narrow content policies for Google3955Search, which cover child sexual abuse material (``CSAM''), highly3956personal information, spam, site owner requests, and valid legal3957requests.3958    We recognize that transparency regarding our own policies and terms3959of service, as well as our interactions with governments, is essential3960to foster public trust and uphold the principles of freedom of3961expression.To this end, YouTube makes its Community Guidelines and3962Terms of Service easily accessible online and publishes data on3963government content removal requests in its biannual Transparency3964Report.39653966    e. In the September 23, 2025 letter, counsel for Alphabet stated:3967``YouTube will provide an opportunity for all creators to rejoin the3968platform if the Company terminated their channels for repeated3969violations of COVID-19 and elections integrity policies that are no3970longer in effect.''39713972    i. To date, how many creators who had their channels terminated3973have had their channels restored?39743975    ii. How many creators does YouTube project are eligible to have3976their channels restored due to termination by YouTube under this3977opportunity to Rejoin?3978    Answer. YouTube had a long standing policy whereby creators whose3979channels had been terminated were permanently banned from the platform.3980We heard frustration from creators that they wanted more options to3981return to YouTube.3982    While our termination policy remains an effective tool to protect3983our community, we recognize that a blanket permanent ban left some3984creators on the sidelines. As a result, we recently rolled out a new3985pilot for a broad set of terminated creators to request a new channel.3986We are considering several factors when evaluating requests for new3987channels, including whether the creator committed particularly severe3988or persistent violations of our Community Guidelines or Terms of3989Service. Creators are also not eligible to apply for a new channel3990until one year after their channel was terminated. However, during the3991year that creators were terminated, they can still appeal the channel3992termination if they think we got it wrong, and YouTube will take3993another look.3994    Relatedly, for Community Guidelines policies that have been3995deprecated--namely COVID-19 and election integrity--we offered creators3996a 30-day grace period to appeal those terminations. Channels were3997reinstated upon appeal.3998    We are happy to brief your office about the status of this pilot3999program.40004001    f. During the hearing, I showed you a chart with an example of a4002video that YouTube blocked in 2022 as part of its censorship of lawful4003content. You admitted during the hearing that the video I showed to you4004on my chart would be considered ``news.'' Why was Google censoring the4005news? Was it wrong to censor the news?4006    Answer. All content on YouTube must abide by our Community4007Guidelines. Each enforcement decision is specific to the content4008policies at issue in the evaluation. Sometimes, content that would4009otherwise violate our Community Guidelines can stay on YouTube when it4010has Educational, Documentary, Scientific, or Artistic (EDSA) context.4011We assess whether content gets an EDSA exception on a case-by-case4012basis--first determining whether there is a Community Guidelines4013violation in the content and, if so, assessing whether there is enough4014context in the content to make an EDSA exception. With respect to the4015chart you are referencing, we would need additional information about4016the specific content at issue to determine whether our enforcement4017action was accurate or made in error. We would be happy to review with4018additional context and information.40194020    Question 3. The European Union's Digital Services Act and the4021United Kingdom's Online Safety Act require social media and online4022search companies to have content moderation policies that ``mitigate''4023illegal content under the laws of EU Member States or the United4024Kingdom, respectively. Like many other foreign jurisdictions, the EU4025and UK tag far more content as illegal than what the First Amendment to4026the United States Constitution would allow.40274028    a. Does Google have an affirmative, written, enacted policy to4029ensure that the First Amendment is the legal standard underlying your4030content moderation policies addressing the expression of United States4031persons?40324033    b. How often are Google's social media or search products,4034policies, and processes tailored to a distinct legal jurisdiction,4035specifically the United States, instead of globally? Why or why not?40364037    c. Please list all instances where a Google product's content4038moderation policy differs in a European jurisdiction from Google's4039policy in the United States and state what the policy difference is.40404041    d. If a foreign jurisdiction provides Google notice that expression4042by a United States person is illegal in that jurisdiction and should be4043suppressed, either through elimination, demotion, or some other means,4044would Google comply with such a notice? If it would depend on4045additional facts or contexts, what facts or contexts?40464047    i. In 2023, 2024, and 2025, how many notices that the expression of4048a United States person is illegal and should be suppressed has Google4049received from a foreign jurisdiction? How many has Google complied4050with? If exact numbers are not available, provide estimates.40514052    ii. When Google does comply with such a notice from a foreign4053jurisdiction, how often does Google suppress that expression only in4054the jurisdiction issuing that notice?40554056    iii. If Google does comply and suppress that expression beyond the4057jurisdiction issuing that notice, how often does Google suppress that4058expression (i) in the United States; or (ii) in other foreign4059jurisdictions beyond the jurisdiction that issued that notice?4060    Answer. Legal standards vary greatly by country and region. Content4061that violates a specific law in one country or region may be legal in4062others. Typically, we remove or restrict access to content only in the4063specific country or region where it is deemed illegal. We have multiple4064tools at our disposal to maximize freedom of expression while complying4065with local laws. For example, we are able to geo-block a video in a4066particular country or region, or, where required, we can use IP4067addresses to IP-block content availability in a country or region.4068    We receive legal demands to remove content through a variety of4069avenues and from authorities at all levels--including via court orders,4070written requests from national and local government agencies, and4071requests from law enforcement professionals. In order for us to4072evaluate a removal request, it must be in writing, as specific as4073possible about the content to be removed, and clear in its explanation4074of how the content is illegal. We do not honor requests that have not4075been made through the appropriate channels, and in cases in which we4076receive verbal requests, we ask for them to be put in writing.4077    We maintain a robust process to receive, evaluate, and act on4078government removal requests. We review these requests closely to4079confirm that they are supported by local laws and to determine whether4080we should remove content as a matter of national law or our platform-4081wide policies.4082    We have a track record of pushing back against overly broad or4083otherwise inappropriate government demands for user data and content4084removals, including objecting to some demands entirely. There are many4085reasons why we may not remove content following a government request.4086Some requests might not be specific enough for us to know what the4087government wants us to remove. In these cases, we ask for more4088information. Other times, we do not take action because the content at4089issue has already been removed by the content owner. As noted above,4090YouTube publishes data on government content removal requests in its4091biannual Transparency Report.4092    In other instances and due to a variety of factors--including4093government demands related to our user data and content policies--we4094have decided to not offer some of our services and platforms in those4095countries. For instance, in mainland China, our core products such as4096Search, Gmail, Maps, and YouTube are not available.40974098    Question 4. On September 23, 2025, Alphabet's counsel sent a letter4099to House Judiciary Committee Chairman Jim Jordan stating that4100``obligations under laws such as the Digital Services Act and Digital4101Markets Act may stifle innovation and restrict access to information.''41024103    a. How do these laws place a disproportionate regulatory burden on4104American Companies?41054106    b. How could these laws pose a risk to freedom of expression4107outside of the European Union, specifically the United States?41084109    c. Alphabet further stated that the DSA ``may open avenues for4110substantive regulation of lawful speech, including through risk4111mitigation, the use of codes of conduct and crisis protocols, and the4112out-of-court dispute settlement mechanisms.'' Please explain further4113how the DSA may open up such avenues for regulation of lawful speech.4114    Answer. An increasing number of countries around the world are4115proposing restrictive laws and content regulations that limit access to4116information and put free expression at risk. Censorship and targeting4117of American companies has long been the case in less democratic4118countries. As these trends have expanded globally, countries are also4119leveraging antitrust actions to exert tighter control over the Internet4120and passing ``hostage-taking laws'' to target or detain workers if they4121do not follow government orders.4122    Increasing censorship and fragmentation will hurt consumers and4123make it more difficult for American businesses to provide cross-border4124services and technologies. Onerous obligations under laws from4125countries across the globe may stifle innovation and restrict access to4126information. We remain mindful of these risks and continue to be4127vigilant in its defense of these rights.4128    Many European regulations and taxes place a disproportionate burden4129on American companies. For example, the European Commission (EC)4130structured the Digital Markets Act (DMA) such that its obligations4131apply only to ``gatekeepers'' that meet certain specific thresholds of4132annual turnover and number of users. These quantitative thresholds are4133set at levels that primarily capture American companies. The EC also4134structured the DMA so that it applies only to gatekeepers that provide4135``core platform services,'' which are defined to exclude sectors (such4136as music streaming, enterprise software, and telecommunications) in4137which European firms are more dominant. Overall, the DMA's thresholds4138capture 22 U.S. products while excluding all European companies and the4139vast majority of Chinese rivals. To date, the EC has only initiated DMA4140investigations against American companies.4141    The DMA is increasingly being deployed to regulate what speech and4142information can be displayed and accessed on American products and4143services. For example, the DMA dictates whether companies can show4144weather forecasts, sports scores, dictionary definitions, stock quotes,4145and other forms of informational content in search results and other4146media. A recent DMA investigation could result in companies being4147forced to display spam, scams, and other low-quality content. In4148addition, the DMA has decreased the visibility of reviews and other4149user-generated content related to restaurants, hotels, and other local4150services. This has resulted in new barriers for people that are seeking4151to access important channels of user expression.4152    In a similar fashion, the DSA was designed as an ``asymmetric''4153regulation that places the heaviest burdens on companies based on size.4154This has had a disproportionate impact on American companies. Among4155other provisions, the regulation includes potential fines of up to six4156percent of a company's global annual turnover; fines are not limited to4157European revenues. DSA regulators also charge large `supervisory fees,'4158also calculated in reference to global turnover, and used to subsidize4159oversight of Chinese and European companies.4160    The Company has also expressed concerns about the risk that the DSA4161may pose to freedom of expression within and outside of the European4162Union, depending on how certain provisions may be enforced. For4163example, DSA risk assessments not only cover systemic risks for the4164dissemination of illegal content, but also cover categories of lawful4165speech considered to pose actual or foreseeable negative effects on4166civic discourse. Additionally, under the DSA, out-of-court dispute4167settlement bodies can be certified to have oversight over both4168notifications of illegal content and application of platforms' global4169policies, including review of platform decisions to keep lawful content4170online in accordance with their policies.4171                                 ______41724173[GRAPHIC(S) NOT AVAILABLE IN TIFF FORMAT]41744175                                 ______41764177   Response to Written Questions Submitted by Hon. Maria Cantwell to4178                            Markham Erickson4179    According to the Pew Research Center, more and more Americans,4180especially young people, get their news from social media platforms,4181including YouTube, Instagram, and TikTok. By design, these platforms4182don't produce news themselves, yet benefit greatly from those that4183generate the content and news that populate their platforms.4184    The rise of these platforms and shifts in the advertising market4185have undermined the economics of local journalism. According to the4186Medill Center's annual report, we have lost more than one-third of our4187newspapers since 2005 and 127 newspapers in the last year alone. And AI4188is only making things worse, with studies showing significantly fewer4189people clicking through to read full articles when an AI summary4190appears. As more consumers turn to AI to find information, newsrooms4191across the country are left producing content that's critical for our4192democracy yet funneled through platforms and models outside of their4193control. Now more than ever, we need to have a system that is4194responsive to the fact that we need reliable information.41954196    Question 1. Does your company have an interest in maintaining the4197economic model for local journalism?4198    Answer. Our core operating principle is to connect people with4199relevant information. A healthy information environment, including news4200and local journalism, is a shared societal asset, and its endurance4201requires a collective effort. Google's work includes developing4202products that help connect Americans with local news content when it is4203relevant and useful to their needs, supporting local journalists with4204tools for data analysis, and helping local media companies to reach an4205audience and monetize their content.4206    Google Search and Google News drive economic value to American news4207businesses by displaying links to relevant news results, connecting4208people to publishers' websites more than 24 billion times per month.4209These website visits provide news publishers with increased readership4210and support publisher revenue through advertising and subscriptions,4211see more information here. According to PwC, each of these visits, on4212average, creates 8 to 10 cents in value for the relevant website owner.4213    Google also pays to license content from local news publishers for4214Google News Showcase. News Showcase is a product feature within Google4215News that gives participating publishers more control over their4216content and how it should appear to our users and, in the US, more than421790 percent of participating publishers are local news sites. In4218Washington state, we partner with a number of local news organizations4219including Lynwood Today, Edmonds News, MLTnews, The Spokesman-Review,4220and Capitol Hill Seattle Blog.4221    Additionally, earlier this year, we launched the AI Sustainability4222Lab, a program to help news publishers explore the use of AI to support4223business objectives. The Seattle Times Affiliate Newspapers was part of4224the first cohort of 15 news outlets to participate, and reported the4225program led to them creating a News Marketing Assistant GPT that lifted4226e-mail referrals by 149 percent.42274228    Question 2. In the era of AI, does Google have a responsibility to4229ensure that local journalism survives?4230    Answer. Google has worked for years to support the long-term4231sustainability of local journalism and recognizes that the preservation4232of a healthy news environment is a shared societal asset, one that4233requires the support of a multi-stakeholder effort including consumers,4234governments, philanthropy organizations and the private sector.4235    At Google, we are committed to doing our part by:42364237   Prioritizing the Web: In the era of AI, more than any other4238        company Google is focused on building products that drive users4239        to directly engage with content creators and publishers'4240        websites. We are focused on fulfilling our part by ensuring our4241        new AI features, like AI Overviews and AI Mode, continue to4242        send valuable traffic to diverse publishers and creators. Our4243        products are distinct from those of some other companies in4244        that they are built to highlight the web, featuring prominent4245        links and visible citation of sources.42464247   Supporting Evolution: We recognize that the news industry is4248        navigating complex challenges, such as ``news deserts.'' We4249        also see growth in digitally-native newspapers and note there4250        are some local news organizations showing growth and success.4251        We aim to work constructively with the industry by developing4252        AI tools that support their work, help them better engage with4253        Americans, and support their financial sustainability.42544255    We stand by our record of support for local news and the news4256ecosystem at large, and believe others in American society could make4257similar contributions.42584259    Question 3. What can we do in the era of AI to make sure that we4260are instilling more competition, growing more diverse media sources,4261and ensuring that quality, reliable information is produced?4262    Answer. In the era of AI, we see a focus on relevance and4263authenticity as being critical to users as they navigate the web, and4264seek to find and engage relevant and helpful content from a range of4265sources. AI has lowered barriers to creation, allowing for a ``long4266tail'' of creativity where independent experts build business models4267around niche topics and authentic voices become more discoverable.4268    We see users increasingly seeking different perspectives and4269opinions, often direct from primary sources and experts, and a growing4270preference for audio and visual formats, in addition to text. Google is4271implementing AI to help meet these needs, enabling users to explore a4272wider range of perspectives relevant to their query, and in a variety4273of formats.4274    Google also recently launched Preferred Sources, a feature that4275provides users with more control over their Search experience by4276enabling them to select the sites they want to see more of, whether4277that is a favorite blog or a local news outlet like the Tacoma News4278Tribune or the Edmonds Beacon.4279    More generally, we continue to invest in and support core4280principles that have long underpinned the utility of Google Search and4281Google News.42824283   Prioritizing Originality: Highlighting websites that create4284        in-depth reporting, unique perspectives, and first-person4285        experiences that go beyond what an AI summary can convey.42864287   Empowering Creators: Providing AI tools like Pinpoint to4288        free up journalists from routine tasks, allowing them to focus4289        on high-impact creative and investigative work.42904291   Ensuring Quality and Trust: Upholding our commitment to4292        quality through sophisticated ranking algorithms and quality4293        control measures that assess content for expertise. We also4294        employ tools like SynthID watermarking of AI-generated imagery4295        and video to provide provenance and context, helping consumers4296        make informed assessments.42974298   Promoting Fair Use: Maintaining balanced copyright4299        frameworks, such as fair use in the United States and flexible4300        text and data mining exceptions in countries like Japan,4301        Singapore, and the European Union is key to combating bias and4302        discrimination in AI systems, preventing skewed outputs, and4303        ensuring there is a balance between the interest of consumers4304        and copyright holders.4305                                 ______43064307      Response to Written Questions Submitted by Hon. Ted Cruz to4308                               Neil Potts4309    Question 1. The European Union's Digital Services Act and the4310United Kingdom's Online Safety Act require social media and online4311search companies to have content moderation policies that ``mitigate''4312illegal content under the laws of EU Member States or the United4313Kingdom, respectively. Like many other foreign jurisdictions, the EU4314and UK tag far more content as illegal than what the First Amendment to4315the United States Constitution would allow.43164317    a. Does Meta have an affirmative, written, enacted policy to ensure4318that the First Amendment is the legal standard underlying your content4319moderation policies addressing the expression of United States persons?4320    Answer. At Meta, we want people to be able to talk openly about the4321issues that matter to them, whether through written comments, photos,4322music, or other means. On platforms where billions of people can have a4323voice, virtually all of the good, bad and ugly is on display. But4324that's free expression, a value at the core of our corporate ethos. We4325appreciate efforts within the Trump Administration and in Congress to4326ensure that international regulatory regimes do not continue to place4327increasingly burdensome and disparate requirements on U.S. companies4328when it comes to moderating content. To that end, we have cooperated4329with a number of related inquiries from Federal regulators and4330lawmakers on Capitol Hill regarding similar topics. In addition, we4331have been able to share lessons from our broader compliance experience4332with other stakeholders in the Administration. This cooperation4333reflects our broader effort to align our operations with the values4334enshrined in the U.S. Constitution, even when operating in4335jurisdictions that do not share, or even recognize, these protections.4336As we have shared, despite our commitment to American values, and in4337some circumstances because of that commitment, we face increasing4338pressure from foreign governments whose legal frameworks often conflict4339with U.S. values and consumer expectations. This can create significant4340operational and legal complexities and raises specific concerns from4341our users about their rights. But we have taken, and will continue to4342take, concrete measures to resist such overly burdensome4343extraterritorial demands, protect American users from foreign4344overreach, and ensure American innovation remains at the forefront of4345global competition.43464347    b. In October 2025, Meta began prohibiting political, electoral,4348and social issue ads on its platforms in the EU in response to the EU's4349Transparency and Targeting of Political Advertising regulation. How4350often are your company's social media or search products, policies, and4351processes tailored to a distinct legal jurisdiction, specifically the4352United States, instead of globally? Why or why not?4353    Answer. With respect to user content, Facebook, Instagram,4354Messenger and Threads are global communities, so our Community4355Standards apply equally to everyone, everywhere and to all types of4356content. Our approach to enforcing these policies may differ depending4357on the jurisdiction, but the rules about what is allowed on our4358platforms remain the same. As one example, Meta last year announced a4359series of significant changes to its content moderation practices4360specifically focused on strengthening free expression in the U.S. Among4361other things, we ended our third-party fact-checking program in the4362United States and moved to a Community Notes model; eliminated a number4363of restrictions on topics that are the subject of frequent political4364discourse and debate; and adopted a more personalized approach to4365political content. In contrast, international regulatory regimes have4366increasingly placed burdensome and disparate requirements on U.S.4367companies, at times even seeking to undermine the core American4368principle of free speech. Despite these challenges, we remain committed4369to freedom of expression globally and have opposed extraterritorial4370legal demands that run contrary to America's fundamental values.4371    Similarly, our Advertising Standards are also global and apply to4372ads running across our services. However, local laws, regulations, and4373where applicable, self-regulatory advertising codes may limit the4374offerings available in different countries. As referenced in your4375question, we no longer allow political, electoral and social issue ads4376on our platforms in the European Union (''EU''). This was a difficult4377decision--one we took in response to the EU's Transparency and4378Targeting of Political Advertising (TTPA) regulation, which introduces4379significant operational challenges and legal uncertainties. Similarly,4380due to highly burdensome regulations, our policies prohibit ads4381targeted to Washington state that relate to Washington's state or local4382elected officials, candidates, elections or ballot initiatives.4383    We continue to believe online political advertising is a vital part4384of modern politics, connecting people to important information about4385the politicians that represent them, and ensuring candidates have a4386cost-effective way of reaching their audiences. That is why Meta has4387gone above and beyond many of our peers--and well beyond what is4388required by law--to ensure the political ads served on our platforms4389are authentic and information about them is transparent.4390    Since 2018, we have had in place tools which provide more4391transparency for ads about politics, elections and social issues than4392any other platform, on or offline, as well as other extensive4393safeguards. Advertisers who run these ads are required to complete an4394authorization process, to prove who they are and where they live, and4395include a ``paid for by'' disclaimer on these ads. These ads are then4396stored in our publicly available Ad Library, where everyone can see4397information about targeting and how much was spent on them.4398    Unfortunately, the TTPA introduces significant, additional4399obligations to our processes and systems that create an untenable level4400of complexity and legal uncertainty for advertisers and platforms4401operating in the EU. For example, the TTPA places extensive4402restrictions on ad targeting and delivery which would restrict how4403political and social issue advertisers can reach their audiences and4404lead to people seeing less relevant ads on our platforms. It is yet4405another threat to the principles of personalized advertising, ignoring4406the benefits to advertisers and the people they want to reach.4407    Despite extensive engagement with policymakers to share these4408concerns, we were left with an impossible choice: alter our services to4409offer an advertising product which does not work for advertisers or4410users, without guarantee that our solution would be viewed as4411compliant, or stop allowing political, electoral and social issue ads4412in the EU. We are not the only company to have been forced into this4413position. Once again, we are seeing regulatory obligations effectively4414remove popular products and services from the market, reducing choice4415and competition.4416    Our decision is specific to the EU. Elsewhere, we will continue to4417provide our industry-leading tools that ensure authentic and4418transparent political advertising. It also will not prevent people in4419the EU from continuing to debate politics on our services, or stop4420politicians, candidates and political office holders from producing and4421sharing political content organically. They just will not be able to4422amplify this through paid advertising.4423    We believe that personalized ads are critical to a wide range of4424advertisers, including those engaged on campaigns to inform voters4425about important social issues that shape public discourse. Regulations,4426like the TTPA, significantly undermine our ability to offer these4427services, not only impacting effectiveness of advertisers' outreach but4428also the ability of voters to access comprehensive information.44294430    c. Please list all instances where Meta's content moderation policy4431differs in a European jurisdiction from Meta's policy in the United4432States and state what the policy difference is.4433    Answer. Please see the response to Question 1.b.44344435    d. If a foreign jurisdiction provides Meta notice that expression4436by a United States person is illegal in that jurisdiction and should be4437suppressed, either through elimination, demotion, or some other means,4438would Meta comply with such a notice? If it would depend on additional4439facts or contexts, what facts or contexts?44404441    i. In 2023, 2024, and 2025, how many notices that the expression of4442a United States person is illegal and should be suppressed has Meta4443received from a foreign jurisdiction? How many has Meta complied with?4444If exact numbers are not available, provide estimates.44454446    ii. When Meta does comply with such a notice, how often does Meta4447suppress that expression only in the jurisdiction issuing that notice?44484449    iii. If Meta does comply and suppresses that expression beyond the4450jurisdiction issuing that notice, how often does Meta suppress that4451expression (i) in the United States; or (ii) in other foreign4452jurisdictions beyond the jurisdiction that issued that notice?4453    Answer. When regulators or government entities believe content on4454our services goes against local law, they may ask us to restrict the4455content. Non-government entities and members of the public may also4456send reports alleging content is unlawful. We may also receive court4457orders. While uncommon, we occasionally receive legal demands that4458assert extraterritorial jurisdiction and request that we restrict the4459availability of content globally. While we respect the law in countries4460where we operate, we strongly oppose any extraterritorial legal demands4461and actively pursue all available options to appeal such orders.4462    When we receive a report or an order, we first review it against4463our Community Standards. If we determine that the content goes against4464our policies, we remove it. If content does not go against our4465policies, in line with our commitments as a member of the Global4466Network Initiative and our Corporate Human Rights Policy, we conduct a4467careful legal review as well as human rights due diligence to determine4468whether the report is valid.4469    As a member of the Global Network Initiative, we are committed to4470narrowly interpreting the jurisdiction of government and law4471enforcement entities and to considering opportunities to push back on4472or challenge requests that are overly broad or inconsistent with4473internal standards.4474    In cases where we believe that reports are not legally valid, are4475overly broad, or are inconsistent with users' rights, we may request4476clarification or take no action.4477    Where we feel we are compelled to act against user content on the4478basis of foreign law rather than our Community Standards, we endeavor4479to restrict access to the content only in the jurisdiction where it is4480alleged to be unlawful and do not impose any other penalties or feature4481restrictions. We also notify the affected user, to the extent possible.4482    Where we are forced to comply with global takedown orders we will4483challenge that decision both before and after the fact, and we publish4484the decisions on our Transparency Center.44854486    Question 2. On August 26, 2024, Meta CEO Mark Zuckerberg sent a4487letter to House Judiciary Chairman Jim Jordan stating, ``I feel4488strongly that we should not compromise our content standards due to4489pressure from any Administration in either direction--and we're ready4490to push back if something like this happens again.'' What processes,4491procedures, or policies has Meta changed since it was pressured by the4492Biden administration in 2021 to ensure Meta is ``ready to push back''4493against government jawboning? Please be specific.4494    Answer. We recently implemented a series of significant changes to4495our content moderation approach in the U.S. designed to bring us back4496to our roots and allow for more speech.4497    As we said at the time of their announcement:44984499   Ending Our Third-Party Fact-Checking Program and Moving to a4500        Community Notes Model. When we launched our independent fact-4501        checking program in 2016, we were very clear that we did not4502        want to be the arbiters of truth. We made what we thought was4503        the best and most reasonable choice at the time, which was to4504        hand that responsibility over to independent fact-checking4505        organizations. The intention of the program was to have these4506        independent experts give people more information about the4507        things they see online, particularly viral hoaxes, so they were4508        able to judge for themselves what they saw and read.45094510    That is not the way things played out, especially in the United4511        States. Experts, like everyone else, have their own biases and4512        perspectives. This showed up in the choices some made about4513        what to fact check and how. Over time we ended up with too much4514        content being fact checked that people would understand to be4515        legitimate political speech and debate. Our system then4516        attached real consequences in the form of intrusive labels and4517        reduced distribution.45184519    So we changed our approach and ended the third-party fact checking4520        program in the United States, moving to a Community Notes4521        program. We have seen this approach work on other platforms--4522        where they empower their community to decide when posts are4523        potentially misleading and need more context, and people across4524        a diverse range of perspectives decide what sort of context is4525        helpful for other users to see. We think this is a better way4526        of achieving our original intention of providing people with4527        information about what they're seeing.45284529    Meta does not write Community Notes or decide which ones show up.4530        They are written and rated by contributing users. Community4531        Notes also require agreement between people with a range of4532        perspectives to help prevent biased ratings. We intend to be4533        transparent about how different viewpoints inform the Notes4534        displayed in our apps and are working on the right way to share4535        this information.45364537    We have phased in Community Notes in the U.S. and continue to4538        improve them. We have stopped demoting fact-checked content,4539        and, instead of overlaying full screen interstitial warnings4540        people have to click through before they can even see a post,4541        we now use a much less obtrusive label indicating that there is4542        additional information for those who want to see it.45434544   Allowing More Speech. We found that we were over-enforcing4545        our rules, limiting legitimate political debate, censoring too4546        much trivial content, and subjecting too many people to4547        frustrating enforcement actions. We want to undo the mission4548        creep that has made our rules too restrictive and too prone to4549        over-enforcement. We got rid of a number of restrictions on4550        topics that are the subject of frequent political discourse and4551        debate.45524553    We also changed our approach to enforcement to reduce the number of4554        mistakes. We historically used automated systems to scan for4555        all policy violations, but this resulted in too many mistakes4556        and too much content being censored that should not have been.4557        So, we have continued focusing these systems on tackling4558        illegal and high-severity violations, like terrorism, child4559        sexual exploitation, drugs, fraud, and scams. For less severe4560        policy violations, we now rely on someone reporting an issue4561        before we take any action.45624563    We also found we were demoting too much content that our systems4564        predicted might violate our standards. We got rid of most of4565        these demotions and now require greater confidence that the4566        content violates for the rest. Additionally, we tuned our4567        systems to require a much higher degree of confidence before a4568        piece of content is taken down.45694570    People are often given the chance to appeal our enforcement4571        decisions and ask us to take another look, but the process can4572        be frustratingly slow and doesn't always get to the right4573        outcome. We added extra staff to this work and, in more cases,4574        we are also now requiring multiple reviewers to reach a4575        determination in order to take something down. We are working4576        on ways to make recovering accounts more straightforward and4577        testing facial recognition technology, and we have started4578        using AI large language models to provide a second opinion on4579        some content before we take enforcement actions.45804581   A Personalized Approach to Political Content. Since 2021, we4582        have made changes to reduce the amount of civic content people4583        see--posts about elections, politics, or social issues--based4584        on the feedback our users gave us that they wanted to see less4585        of this content. But this was a pretty blunt approach. We4586        started phasing this back into Facebook, Instagram, and Threads4587        with a more personalized approach so that people who want to4588        see more political content in their feeds can.45894590    We are also continually testing how we deliver personalized4591        experiences and have conducted testing around civic content. As4592        a result, we started treating civic content from people and4593        Pages users follow on Facebook more like any other content in4594        their feed, and we started ranking and showing people that4595        content based on explicit signals (for example, liking a piece4596        of content) and implicit signals (like viewing posts) that help4597        us predict what is meaningful to people. We are also4598        recommending more political content based on these personalized4599        signals and are expanding the options people have to control4600        how much of this content they see.46014602    We believe these changes are having a positive impact. As reflected4603in our Community Standards Enforcement Report for the first quarter of46042025, we saw a roughly 50 percent reduction in enforcement mistakes on4605our platforms in the United States from the last quarter of 2024 to the4606first quarter of 2025. During that same time period, the low prevalence4607of violating content largely remained unchanged for most problem areas.4608This improvement follows the earlier commitment we made to change our4609focus to proactively enforcing high-severity violations and enhancing4610our accuracy through system audits and additional signals, as described4611above.4612                                 ______46134614   Response to Written Questions Submitted by Hon. Maria Cantwell to4615                               Neil Potts4616    Local Journalism. According to the Pew Research Center, more and4617more Americans, especially young people, get their news from social4618media platforms, including YouTube, Instagram, and TikTok. By design,4619these platforms don't produce news themselves yet benefit greatly from4620those that generate the content and news that populate their platforms.4621    The rise of these platforms and shifts in the advertising market4622have undermined the economics of local journalism. According to the4623Medill Center's annual report, we have lost more than one-third of our4624newspapers since 2005 and 127 newspapers in the last year alone. And AI4625is only making things worse, with studies showing significantly fewer4626people clicking through to read full articles when an AI summary4627appears. As more consumers turn to AI to find information, newsrooms4628across the country are left producing content that's critical for our4629democracy yet funneled through platforms and models outside of their4630control. Now more than ever, we need to have a system that is4631responsive to the fact that we need reliable information.46324633    Question 1. Does your company have an interest in maintaining the4634economic model for local journalism?4635    Answer. We want to serve all our users, including local4636journalists, and our free platforms and low-cost advertising provide4637economic benefit to news outlets. Publishers choose to post their4638content on our platforms to expand their reach, gain new audiences and4639increase their revenue. The links that publishers share on Facebook4640drive new traffic to their own sites, and publishers keep 100 percent4641of the revenue from traffic and subscriptions derived from these4642outbound links from Facebook. Many local journalists have highlighted4643the importance of this. These publishers typically choose to put links4644on our site.46454646    Question 2. In the era of AI, does Meta have a responsibility to4647ensure that local journalism survives?4648    Answer. Posts with links to news are a small part of the Facebook4649experience for the majority of people who use the platform. We invest4650in our services to connect people to diverse sources of information4651that educate and entertain them. These investments serve communities on4652and off our platforms. We are proud that news publishers derive4653significant value from our platforms, which are free and have tools4654that derive unique value to publishers and journalists of all sizes,4655helping them build sustainable business models. This is why so many4656publishers voluntarily post their own content to our platforms with4657links to their news stories. That engagement links directly back to the4658publisher's site and generates revenue opportunities for those4659publishers.46604661    Question 3. What can we do in the era of AI to make sure that we4662are instilling more competition, growing more diverse media sources,4663and ensuring that quality, reliable information is produced?4664    Answer. AI offers an opportunity to expand access to information4665and help ensure a vibrant, diverse information ecosystem. AI is poised4666to drive innovation, enhance human productivity, and inject trillions4667of dollars into the global economy.4668    One example is open-source AI models, like Llama, which empower4669organizations to use and build upon them for free. Open-source AI4670models allow businesses of all sizes, including news organizations, to4671create innovative new products and tools that benefit individuals,4672society and the economy--saving time and money in the process. The4673Washington Post, for example, launched ``Ask The Post,'' an AI tool4674built with Llama that helps users find answers from its published4675reporting, in an effort to democratize access to information and meet4676audiences where they are. Readers can ask the chatbot questions and4677receive factual responses based on information from The Washington4678Post's article archives dating back to 2016. ``Ask The Post'' responds4679in the newspaper's voice and links to source articles, helping to4680ensure transparency and accountability. The Washington Post was able to4681do this without the limitations and cost restrictions of other4682proprietary models.4683                                 ______46844685      Response to Written Questions Submitted by Hon. Ted Cruz to4686                              Will Creeley4687    Question 1. Does the First Amendment protect the information4688Americans are permitted to receive and not just the content they can4689express?4690    Answer. Yes, the First Amendment protects our right to receive4691information. In decisions dating back decades, the Supreme Court of the4692United States has consistently recognized that the same First Amendment4693freedom that protects our right to share an idea ``necessarily protects4694the right to receive it.'' Martin v. Struthers, 319 U.S. 141, 1434695(1943); see also Thomas v. Collins, 323 U.S. 516, 534 (1945) (striking4696down statutory restriction on labor leader's speech as violative of4697both his ``right to speak and the rights of the workers to hear what he4698had to say''); Stanley v. Georgia, 394 U.S. 557, 564 (1969) (``It is4699now well established that the Constitution protects the right to4700receive information and ideas.''); Va. State Bd. of Pharmacy v. Va.4701Citizens Consumer Council, 425 U.S. 748, 756 (1976) (First Amendment4702protects a speaker's ``communication, to its source and to its4703recipients both.''). As Justice Brennan put it: ``The dissemination of4704ideas can accomplish nothing if otherwise willing addressees are not4705free to receive and consider them. It would be a barren marketplace of4706ideas that had only sellers and no buyers.'' Lamont v. Postmaster Gen.,4707381 U.S. 301, 308 (1965) (Brennan, J., concurring). The bottom line is4708that the First Amendment generally prohibits government restrictions on4709willing speakers reaching willing listeners, or interference with what4710speakers choose to say.4711    But while the First Amendment bars the government from interfering4712with our right to receive information, when a private entity hosts4713speech--as with newspaper opinion pages, for example, or social media4714posts--that platform possesses its own First Amendment right to decide4715for itself what content it makes available, and the government may not4716compel it to choose otherwise. That's why the Supreme Court has4717consistently ``barred the government from forcing a private speaker to4718present views it wished to spurn in order to rejigger the expressive4719realm.'' Moody v. NetChoice, LLC, 603 U.S. 707, 727-733 (2024)4720(reviewing cases). ``However imperfect the private marketplace of4721ideas,'' the First Amendment prohibits the ``worse proposal--the4722government itself deciding when speech was imbalanced, and then4723coercing speakers to provide more of some views or less of others.''4724Id. at 733.47254726    Question 2. When, if ever, does the First Amendment apply within4727the context of artificial intelligence?4728    Answer. Always, potentially. Artificial intelligence generally4729involves and/or operates through expression. And as with any other4730technology, when artificial intelligence is used for expressive4731purposes, the First Amendment applies. People, not the tools we use to4732communicate, possess First Amendment rights--and artificial4733intelligence is a new tool, one with potentially powerful ramifications4734for our ability to generate and exchange information and ideas.4735    Technological advances in how we communicate with one another do4736not alter or diminish the First Amendment's protection against4737government censorship. When Americans use artificial intelligence to4738invite, facilitate, and/or produce expression, the First Amendment4739applies just as it does when we share ideas and information via the4740printed page, radio, television, telephone, or internet. The First4741Amendment's protection remains constant, as Justice Scalia recognized4742in a case about the constitutional implications of government4743restrictions on video games: ``[W]hatever the challenges of applying4744the Constitution to ever-advancing technology, `the basic principles of4745freedom of speech and the press, like the First Amendment's command, do4746not vary' when a new and different medium for communication appears.''4747Brown v. Entm't Merchs. Ass'n, 564 U.S. 786, 790 (2011) (quoting Joseph4748Burstyn, Inc. v. Wilson, 343 U.S. 495, 503 (1952)). Accordingly, using4749artificial intelligence to create, disseminate, and receive information4750is protected by the First Amendment.47514752    Question 3. What are the First Amendment and free speech4753implications of government attempts to alter content generated by4754artificial intelligence to address alleged ``bias'' or to pressure4755artificial intelligence companies to set up processes to moderate or4756eliminate alleged ``disinformation,'' as well as ``offensive'' or4757``objectionable'' content? Do such efforts run afoul of the First4758Amendment? How?4759    Answer. Each of these efforts would violate well-established4760Supreme Court precedent and rob Americans of the right to engage in4761open discourse free from government censorship. The First Amendment4762prohibits the government from defining and prohibiting ``bias'' or4763``disinformation,'' or ``offensive'' or ``objectionable'' speech,4764because any such restriction would require the government to dictate4765answers to inherently subjective determinations. Because ``one man's4766vulgarity is another's lyric,'' our First Amendment jurisprudence4767rightfully recognizes ``governmental officials cannot make principled4768distinctions'' between the two. Cohen v. California, 403 U.S. 15, 254769(1971). Instead, the First Amendment requires that these subjective4770determinations be left where they belong: with the individual.4771    The First Amendment bars the government from altering or4772restricting private speech--including expressive content generated by4773artificial intelligence--to suppress ``biased'' viewpoints. ``On the4774spectrum of dangers to free expression, there are few greater than4775allowing the government to change the speech of private actors in order4776to achieve its own conception of speech nirvana.''4777    Moody v. NetChoice, LLC, 603 U.S. 707, 741-42 (2024). That's4778because any such effort would violate the First Amendment in several4779fundamental respects. Perhaps most predominantly, the First Amendment4780prohibits viewpoint discrimination. ``It is axiomatic that the4781government may not regulate speech based on its substantive content or4782the message it conveys,'' Rosenberger v. Rector & Visitors of the Univ.4783of Va., 515 U.S. 819, 828 (1995), because allowing the state to favor4784some views while restricting others would render freedom of expression4785illusory. The Founders understood the First Amendment to bar precisely4786this form of state repression. In our system, as James Madison4787articulated, ``the censorial power is in the people over the4788government, and not in the government over the people.'' 4 Annals of4789Cong. 934 (1794).4790    A government-imposed burden or ban on ``biased'' expression4791generated by artificial intelligence would also be impermissibly vague,4792thus violating both the First Amendment and the Fifth Amendment's Due4793Process clause. Because determinations of bias are necessarily4794subjective, a restriction on biased expression would fail to afford4795``fair warning'' to artificial intelligence's developers, distributors,4796and users, depriving them of the notice due process requires and4797allowing for ``arbitrary and discriminatory enforcement.'' Grayned v.4798City of Rockford, 408 U.S. 104, 108-09 (1972).4799    Vagueness is particularly problematic in the First Amendment4800context, because ``[u]ncertain meanings inevitably lead citizens to4801`steer far wider of the unlawful zone' . . . than if the boundaries of4802the forbidden areas were clearly marked.' '' Id. To prevent that4803chilling effect, laws and regulations that may impact speech protected4804by the First Amendment ``demands a greater degree of specificity than4805in other contexts.'' Smith v. Goguen, 415 U.S. 566, 573 (1974). Our4806Bill of Rights ``requires the invalidation of laws that are4807impermissibly vague''--and because a broad ban on ``bias'' would be4808``so standardless that it authorizes or encourages seriously4809discriminatory enforcement,'' it would be unconstitutional. FCC v. Fox4810Television Stations, Inc., 567 U.S. 239, 253 (2012) (citing United4811States v. Williams, 553 U.S. 285, 306 (2008)).4812    Government attempts to ``pressure artificial intelligence companies4813to set up processes to moderate or eliminate alleged `disinformation,'4814as well as `offensive' or `objectionable' content'' would fare no4815better. Like a ban on ``bias,'' these operative terms--4816``disinformation,'' or ``offensive'' or ``objectionable''--are both4817viewpoint discriminatory and inherently subjective, and thus fail to4818pass constitutional muster. Further, the government cannot require4819private companies to moderate speech in a particular way. Moody, 6034820U.S. at 734 (``The government may not, in supposed pursuit of better4821expressive balance, alter a private speaker's own editorial choices4822about the mix of speech it wants to convey.'').4823    These restrictions on government regulation of speech apply4824regardless of whether the regulation comes via a formal edict or4825informal coercion. Compounding the constitutional problem, the4826government cannot ``pressure'' or coerce private companies to censor4827speech; as noted in my opening testimony, indirect or informal attempts4828to censor speech in this way violate the First Amendment as surely as4829direct censorship. As the Supreme Court reaffirmed just last year: ``A4830government official cannot coerce a private party to punish or suppress4831disfavored speech on her behalf.'' NRA of Am. v. Vullo, 602 U.S. 175,4832190 (2024).4833    Our jurisprudence recognizes that speech is ``powerful''--that it4834``can stir people to action, move them to tears of both joy and sorrow,4835and . . . inflict great pain.'' Snyder v. Phelps, 562 U.S. 443, 460-614836(2011). The temptation to regulate that power in an attempt to prevent4837that pain is evergreen. But ``[a]s a Nation we have chosen a different4838course--to protect even hurtful speech on public issues to ensure that4839we do not stifle public debate.'' Id. No matter how well-intentioned,4840any governmental effort to restrict ``hurtful speech''--here,4841restrictions on content generated by artificial intelligence--would4842ultimately erode our ability to decide for ourselves what is good and4843true to ``permit the continued building of our politics and culture,4844and to assure self-fulfillment for each individual.'' Police Dep't of4845Chi. v. Mosley, 408 U.S. 92, 95-96 (1972).48464847    Question 4. What should I keep in mind as I craft legislation to4848ensure that Americans are fully protected against government efforts to4849decide or limit what they can read, particularly in the context of4850artificial intelligence?4851    Answer. We must defend our First Amendment right to use artificial4852intelligence to freely express ourselves--and that defense becomes all4853the more essential if it does, in fact, revolutionize our ability to4854communicate with each other and to generate and share knowledge. To4855that end, FIRE urges lawmakers to remember two central points as they4856consider legislation regarding artificial intelligence.48574858  (1)  Artificial intelligence is in its infancy. We are just beginning4859        to understand its possibilities. So imposing broadly4860        prophylactic laws or regulations now to restrict its4861        development and expressive uses risks foreclosing potentially4862        invaluable applications. To the extent that artificial4863        intelligence creates particular policy problems in time,4864        legislators and regulators may act within the bounds of the4865        Constitution to address them. But preemptory action driven by4866        fear about how the technology may impact us will impede4867        innovation. The immense power of the Internet to democratize4868        speech was unleashed in large part because Congress acted4869        quickly to prevent it from being stifled by frivolous4870        litigation or censorial regulations early in its development.4871        We should remember that lesson now.48724873  (2)  Some policymakers fear the use of artificial intelligence will4874        result in a variety of social harms--facilitating4875        discrimination in contexts like medical care, employment, or4876        housing, for example, or generating ``deepfakes'' so lifelike4877        as to cause social and political consequences. It is true that4878        artificial intelligence can be wielded, like any tool, to4879        antisocial and unlawful ends. But it is also true that existing4880        law already prohibits and provides legal redress for the vast4881        majority of the misconduct driving legislators' concerns. We do4882        not enact a new First Amendment to accompany each new4883        technology. We should not enact new civil and criminal codes,4884        either.48854886    And because artificial intelligence companies are just as4887susceptible to coercive governmental pressure as social media4888platforms, both legislators and the public should be vigilant against4889policymakers and regulators jawboning developers into changing models'4890outputs to accord with their political or ideological preferences.4891Pressure campaigns are already underway.\1\4892---------------------------------------------------------------------------4893    \1\ See, e.g., Press Release, Off. of Att'y Gen. of Mo., Attorney4894General Bailey Fights To Expose Big Tech Censorship of President Trump4895as AI Chatbots Produce Fake News (July 9, 2025), https://ago.mo.gov/4896attorney-general-bailey-fights-to-expose-big-tech-censorship-of-4897president-trump-as-ai-chatbots-produce-fake-news (``Today, Missouri4898Attorney General Andrew Bailey sent a formal demand letter to Google,4899Microsoft, OpenAI, and Meta regarding biased and factually inaccurate4900responses produced by the companies' artificial intelligence4901chatbots.'').4902---------------------------------------------------------------------------4903                                 ______49044905   Response to Written Questions Submitted by Hon. Maria Cantwell to4906                              Will Creeley4907    Platform Settlements. President Trump sued the two social media4908platforms testifying at this hearing because they removed him from4909their platforms after the January 6th insurrection.4910    Despite the fact that legal experts called these lawsuits4911meritless, both Meta and Google settled their lawsuits this year, once4912President Trump was back in office.49134914    Question 1. Mr. Creeley, you have called this strategy of suing4915media companies ``dictatorial.'' Can you explain why the President is4916``dictatorial'' when he sues media organizations over their editorial4917decisions?4918    Answer. Dictators abuse power to silence dissent, control the4919narrative, and coerce targets into both immediate and anticipatory4920obedience. They ignore or abolish legal limits on their authority. By4921flouting longstanding First Amendment precedent to file legally4922baseless lawsuits, then abusing his authority to pressure them into4923settlements, President Trump is checking those boxes.4924    Filing as a private citizen, Trump's 2021 suits against Meta and4925Google alleged that the companies violated his First Amendment rights4926by barring him from their platforms and removing his content. Trump4927argued Meta and Google were ``state actors'' based on their4928interactions with the government, but his filings relied on erroneous4929understandings of the law. More centrally, the First Amendment protects4930private entities' exercise of editorial discretion, including the4931decisions of private companies like Meta and Google to moderate content4932on their sites. This is blackletter First Amendment law, reaffirmed by4933the Supreme Court just last year. Moody v. NetChoice, LLC, 603 U.S. 7074934(2024). The Court has ``time and again held'' that attempts to ``alter4935[speakers'] choices about the views they will, and will not, convey . .4936. interfere with protected speech.'' Id. at 737-38. So legal4937commentators correctly described the lawsuits as baseless, as did4938Google and Meta themselves.4939    After Trump won re-election, however, the lawsuits' lack of legal4940merit apparently became less consequential. Rather than continue to4941contest now-President Trump's claims and risk his ire, the companies4942settled--as have other media entities, private universities, and law4943firms. In each instance, the determinative calculations appear to have4944focused more on political and financial consequences rather than legal4945merit, given the strength of the First Amendment arguments available to4946each targeted entity against President Trump's plain abuses of4947power.\2\ These transactional capitulations are a result of the fear4948engendered by the president's demonstrated willingness to abuse the4949power of his office and those of his political appointees to punish4950people and entities whose speech he opposes.4951---------------------------------------------------------------------------4952    \2\ See, e.g., Angel Eduardo, Why YouTube caving to Trump is4953cowardly, Foundation for Individual Rights and Expression (Oct. 2,49542025) (``What makes all this worse is that the lawsuits are based on4955obviously meritless claims that would never withstand scrutiny if they4956actually went to court.'').49574958    Trump Media Lawsuits. In addition to his actions against the4959platforms, President Trump has filed a series of lawsuits against media4960organizations. These include a lawsuit against CBS for editing a 604961Minutes interview with his opponent in the 2024 election, Vice4962President Harris; a lawsuit against the New York Times for defamation4963based on reporting about his time hosting The Apprentice; and even a4964lawsuit against the Des Moines Register and its pollster Ann Selzer for4965publishing a poll showing him losing Iowa in 2024. While some media4966companies have fought these lawsuits, others have settled with the4967---------------------------------------------------------------------------4968President.49694970    Question 1. Mr. Creeley, what is the effect of these lawsuits on4971the free press?4972    Answer. President Trump's lawsuits are a cynical effort to chill4973critical coverage by bullying media outlets into silence. The4974president's suits flatly ignore our ``profound national commitment to4975the principle that debate on public issues should be uninhibited,4976robust, and wide-open.'' N.Y. Times Co. v. Sullivan, 376 U.S. 254, 2704977(1964). Instead, they aim to impose prohibitive costs on media outlets4978that dare to publish and broadcast news and views the president doesn't4979like.4980    The suits rely on bizarre legal theories that would be laughable4981were they not so dangerous. For example, the president's lawsuit4982against our client, the pollster J. Ann Selzer, claims that her poll4983constituted ``fraud'' under state law. But as my colleagues have4984explained at length in our filings, the president's argument has no4985support in law. Selzer's poll was entirely protected by the First4986Amendment; getting a prediction wrong doesn't render a poll ``fraud.''4987Earlier this month, a Federal court agreed, dismissing with prejudice a4988copycat lawsuit filed by a Des Moines Register subscriber that closely4989tracked the president's claims. Donnelly v. Des Moines Register & Trib.4990Co., No. 4:25-cv-00150-RGE-WPK (S.D. Iowa Nov. 6, 2025) (order granting4991defendants' motions to dismiss). As the court explained, ``there is no4992free pass around the First Amendment.'' Id. at *6.4993    Lawsuits like the one against Selzer are classic examples of4994``SLAPPs'': strategic lawsuits against public participation, designed4995to chill speech and punish speakers through costly litigation. That4996these suits are without merit is beside the point; forcing opponents4997into lengthy court battles is the goal. To prevent vexatious SLAPP4998litigants from silencing speech, 38 states (and the District of4999Columbia) to date have enacted legislation allowing for early dismissal5000of meritless suits.5001    But unlike a typical plaintiff, the President's suits present a5002uniquely pernicious threat because of the power of his office--and his5003willingness to abuse it to silence speech. The president has brazenly5004sought to leverage the power of Federal regulatory agencies to force5005favorable resolutions of his lawsuits against media outlets.5006    Take President Trump's lawsuit against CBS, for example. As in his5007suit against Selzer, the President baselessly alleged protected5008journalism--here, CBS' routine edits of a 60 Minutes interview with5009Vice President Kamala Harris--is somehow consumer fraud. It isn't. But5010Federal Communication Commission Chairman Brendan Carr exerted5011regulatory pressure on CBS, reinstating a previously dismissed5012investigation of the network for alleged ``news distortion'' and5013stalling approval of a proposed merger between CBS corporate parent5014Paramount Global and Skydance Media.5015    As my colleagues and I pointed out in FIRE's public comment to the5016agency, Carr's pressure campaign constituted jawboning in broad5017daylight:50185019        The purpose and timing of this inquiry are both obvious and5020        unjustifiable. Launching a politically fraught investigation5021        based on such a paper-thin complaint in these circumstances is5022        alone a compelling example of regulatory abuse. But to5023        resurrect the flimsy complaint after it was fully and properly5024        interred by staff dismissal, and to do so in support of the5025        President's private litigation position, is all but a signed5026        confession of unconstitutional jawboning. The Commission can5027        begin to recover some dignity only by dropping the matter5028        immediately.\3\5029---------------------------------------------------------------------------5030    \3\ A copy of our comment was included with my written testimony5031and is available online at https://www.thefire.org/sites/default/files/50322025/03/FIRE%20Comments%20on%20FCC%205033News%20Distortion%20Complaint.pdf.50345035    And the agency's narrow, tightly cabined ``news distortion''5036doctrine does not and cannot overcome the First Amendment's bar against5037the FCC or its Chairman exerting power over a broadcaster's editorial5038decisions.5039    Nevertheless, CBS settled for $16 million, and the FCC approved the5040merger--complete with a new ``ombudsman,'' Kenneth Weinstein,5041previously President Trump's nominee for ambassador to Japan. According5042to Carr, Weinstein's job will include ensuring CBS airs of ``a5043diversity of viewpoints from across the political and ideological5044spectrum'' in its programming.\4\ A range of views may or may not make5045for compelling television. But that must be the choice of the5046broadcaster, not the government.5047---------------------------------------------------------------------------5048    \4\ David Folkenflik, CBS shifts to appease the right under new5049owner, NPR (Sept. 12, 2025), https://www.npr.org/2025/09/12/nx-s1-50505537152/cbs-news-ellison-steps-appease-trump.5051---------------------------------------------------------------------------5052    The full impact of the President's litigation isn't easy to assess.5053It's hard to count stories that have been spiked, broadcasts that never5054made the air, and assignments that have been revoked. But it is plain5055that the President's willingness to abuse both the legal system, by5056filing meritless lawsuits against media outlets, and the power of his5057office, by invoking regulatory authority to pressure outlets to bow to5058his political worldview, has made publishing and broadcasting views he5059doesn't like far riskier. That makes our free press less so, and we all5060suffer as a result.50615062    FCC Censorship. There has been a lot of attention--and rightly so--5063on FCC Chairman Brendan Carr's threats to knock Jimmy Kimmel off the5064air. But Carr's censorship of Kimmel is just one in a series of attacks5065on the First Amendment. Although the FCC is supposed to be independent,5066under Brendan Carr, the agency has become the President's speech5067police.5068    Since taking over the FCC, Carr has reopened frivolous inquiries5069into CBS, NBC, and ABC about their coverage of the 2024 election;5070launched investigations into NPR and PBS stations; and publicly5071threatened broadcast TV and radio stations based on the content of5072their coverage of immigration issues.50735074    Question 1. Mr. Creeley, does the FCC have authority to dictate5075coverage decisions to journalists?5076    Answer. Absolutely not. Per the First Amendment, Federal law, and5077longstanding precedent, the FCC has no business dictating the editorial5078choices of media outlets.5079    The Communications Act is clear: The FCC cannot exercise ``the5080power of censorship,'' nor impose any ``regulation or condition . . .5081which shall interfere with the right of free speech by means of radio5082communication.'' 47 U.S.C. Sec. 326. For more than fifty years, the5083Commission understood that it ``is not the national arbiter of the5084truth.'' Complaints Covering CBS Program ``Hunger in America,'' 205085F.C.C.2d 143, 151 (1969). Likewise, the Supreme Court has rejected as5086unconstitutional FCC action that risks the ``substantial abridgment of5087important journalistic freedoms which the First Amendment jealously5088protects.'' FCC v League of Women Voters of Cal., 468 U.S. 364, 4025089(1984).5090    In the United States, `` `editorial control and judgment,' not5091official decree,'' determine a broadcast's ``choice of material.''5092Miami Herald Pub. Co., Div. of Knight Newspapers, Inc. v. Tornillo, 4185093U.S. 241, 258 (1974). Government actors cannot dictate to journalists,5094editors, or broadcasters what to say or air; ``any such a compulsion to5095publish that which `reason' tells them should not be published is5096unconstitutional.'' Id. (citation and internal quotation omitted).5097    As a Commissioner, before taking the reins of the agency, Chairman5098Carr once understood that ``[a] newsroom's decision about what stories5099to cover and how to frame them should be beyond the reach of any5100government official, not targeted by them.'' \5\ FIRE will continue to5101remind him of what he has since forgotten.5102---------------------------------------------------------------------------5103    \5\ Press Release, FCC, FCC Commissioner Carr Responds to5104Democrats' Efforts to Censor Newsrooms (Feb. 22, 2021), https://5105docs.fcc.gov/public/attachments/DOC-370165A1.pdf.51065107    Attacks on the First Amendment. Congress created the Federal Trade5108Commission to protect consumers from unfair and deceptive practices and5109ensure a competitive free market. FTC Chairman Andrew Ferguson,5110however, has been using the FTC's authority to bully social media5111platforms into changing their content moderation decisions to favor5112conservative content and force companies to advertise on Elon Musk's X5113platform.5114    This is despite the fact that the Supreme Court has recognized that5115social media platforms have First Amendment rights to ``select and5116shape other parties' expression into their own curated speech5117products.''51185119    Question 1. Mr. Creeley, is Chairman Ferguson's campaign consistent5120with the First Amendment?5121    Answer. No. The First Amendment protects the editorial decisions of5122social media companies just as it protects the editorial decisions of5123newspapers or broadcasters. ``The principle does not change because the5124curated compilation has gone from the physical to the virtual world. In5125the latter, as in the former, government efforts to alter an edited5126compilation of third-party expression are subject to judicial review5127for compliance with the First Amendment.'' Moody v. NetChoice, LLC, 6035128U.S. 707, 717 (2024). And the First Amendment protects social media5129companies and other private entities from all governmental efforts to5130restrict their expressive rights, whether the threat comes from state5131or Federal law, as in Moody, or via pressure from agency action. See,5132e.g., NRA of Am. v. Vullo, 602 U.S. 175 (2024) (director of state5133regulatory agency likely violated First Amendment by pressuring5134regulated entities to end providing services to National Rifle5135Association because of its advocacy).5136    Relatedly, FIRE has opposed Chairman Ferguson's use of sweeping5137civil investigative demands to retaliate against Media Matters for5138America's research and reporting about X Corp. and Elon Musk. See Br.5139Amici Curiae FIRE Supp. Plaintiff-Appellee, Media Matters for Am. v.5140FTC, No. 25-5302 (D.C. Cir. Oct. 23, 2025). Last month, the United5141States Court of Appeals for the District of Columbia Circuit denied the5142Commission's motion to stay a Federal district court's preliminary5143injunction against the investigative demands, noting among other5144factors Chairman Ferguson's statements regarding his desire to5145investigate Media Matters for its viewpoints. Media Matters for Am. v.5146FTC, No. 25-5302, at *21 (D.C. Cir. Oct. 23, 2025).51475148    Press Walkout from Pentagon. Earlier this month, dozens of5149journalists--including those from conservative outlets like Fox News5150and Newsmax--turned in their press badges and walked out of the5151Pentagon rather than agree to a new policy that gave the Defense5152Department the ability to control the information they solicit or5153obtain. Last week, the Pentagon announced a new group of media outlets5154that agreed to its unprecedented restrictions on reporting.51555156    Question 1. Mr. Creeley, is the Pentagon's new press policy5157consistent with the First Amendment?5158    Answer. No. The Pentagon's new press policy runs afoul of the First5159Amendment. The refusal of news organizations to submit to the policy is5160justified and commendable.5161    Courts have long recognized that ``a major purpose of that5162Amendment was to protect the free discussion of governmental affairs.''5163Mills v. Alabama, 384 U.S. 214, 218 (1966). The Founders thus5164``specifically selected the press'' for constitutional protection as5165``a powerful antidote to any abuses of power by governmental officials5166and as a constitutionally chosen means for keeping officials elected by5167the people responsible to all the people whom they were selected to5168serve.'' Id. at 219. But the Pentagon's new press policy contains vague5169restrictions that infringe upon these long-established First Amendment5170principles.5171    Perhaps most troublingly, the policy attempts to draw a distinction5172between ``lawfully requesting information from the government and5173actively soliciting or encouraging government employees to break the5174law,'' contending that ``[t]he First Amendment does not permit5175journalists to solicit government employees to violate the law by5176providing confidential government information.'' But that contention5177has no support in the law. ``The right of citizens to inquire, to hear,5178to speak, and to use information'' is protected by the First Amendment.5179Citizens United v. FEC, 558 U.S. 310, 339, 349 (2010). That includes5180newsgathering. The press possesses an ``undoubted right to gather news5181`from any source by means within the law.' '' Houchins v. KQED, Inc.,5182438 U.S. 1, 11 (1978) (quoting Branzburg v. Hayes, 408 U.S. 665, 681-825183(1972)). And the First Amendment protects reporters who gain5184information from officials ``simply by asking.'' Smith v. Daily Mail5185Publ'g Co., 443 U.S. 97, 99, 103-04 (1979).5186    The policy also presents a practical problem. When reporters5187exercise their First Amendment right to ask questions of government5188officials, they don't determine the answers they may receive. By5189targeting reporters, instead of government officials who may provide5190confidential information, the policy misplaces the burden and restricts5191protected expressive activity. The policy's vague terms also empower5192the government to punish reporters simply for doing their job: asking5193questions.51945195    Brendan Carr Jawboning: Senator Cruz's memo laments that the Biden5196administration had tried to ``shut down the speech of Americans with5197whom it disagreed.'' Most recently, Chairman Carr threatened ABC5198affiliate broadcast licenses over remarks Jimmy Kimmel made on the air.51995200    Question 1. Is Chairman Carr trying to ``shut down the speech of5201Americans with whom [he] disagree[s]'' and does it matter, for purposes5202of the First Amendment, if Jimmy Kimmel's speech was offensive?52035204    Question 2. Does it matter, for purposes of the First Amendment, if5205Jimmy Kimmel's speech was offensive? Does it matter, for purposes of5206the First Amendment, that Jimmy Kimmel's show is not pure news, but5207also has a comedy component?5208    Answers. Yes, Chairman Carr was trying to shut down the speech of5209Americans with who he disagrees, if we take him at his word. Chairman5210Carr has been clear about targeting broadcasters and programming,5211including Kimmel and Disney, because of his disagreement with the views5212they express. After Kimmel's monologue on Charlie Kirk's alleged5213assassin, for example, Carr said the comments created a ``very, very5214serious issue right now for Disney.'' And just this week, Chairman Carr5215reposted President Trump's call for NBC to fire late night host Seth5216Meyers because, in the President's words, he ``is suffering from an5217incurable case of Trump Derangement Syndrome.'' \6\ But5218---------------------------------------------------------------------------5219    \6\ Brendan Carr (@BrendanCarrFCC), X (Nov. 15, 2025, 7:28 PM),5220https://x.com/Brendan5221CarrFCC/status/1989853016174866891?s=20.5222---------------------------------------------------------------------------5223    And no, it does not matter if some, many, most, or even all5224Americans found Kimmel's speech offensive. The First Amendment protects5225offensive speech and prohibits the government from discriminating on5226the basis of viewpoint. ``Giving offense is a viewpoint. The `public5227expression of ideas may not be prohibited merely because the ideas are5228themselves offensive to some of their hearers.' '' Matal v. Tam, 5825229U.S. 218, 220 (2017) (quoting Street v. New York, 394 U.S. 576, 5925230(1969).5231    Nor does it matter, from a First Amendment standpoint, that5232Kimmel's show mixes comedy and news. Whether a show features comedy,5233news, or a blend of both, the First Amendment bars the FCC from5234supplanting the broadcaster's editorial judgments with its own. Turner5235Broad. Sys. v. FCC, 512 U.S. 622, 650 (1994) (``The FCC is forbidden by5236statute from engaging in `censorship' or from promulgating any5237regulation `which shall interfere with the [broadcasters'] right of5238free speech.' '') (quoting 47 U.S.C. Sec. 326). See also Network5239Coverage of the Democratic Nat'l Convention, 16 F.C.C.2d 650, 657-585240(1969) (reaffirming the agency's ``general rule . . . that we do not5241sit to review the broadcaster's news judgment, the quality of his news5242and public affairs reporting, or his taste.'').5243                                 ______52445245   Response to Written Questions Submitted by Hon. Maria Cantwell to5246                              Harold Feld5247    Local News and Media Consolidation. Americans rely on local news to5248understand what's going on in their communities. According to the Pew5249Research Center, nearly three-quarters of Americans trust their local5250newspapers and broadcasters. But local news faces serious headwinds.5251Recently, the Medill Center reported that 50 million Americans have no5252access to local news, a new high.5253    The advertising-supported model of local newspapers and local5254broadcast has declined, with an increasing share of revenue going to5255tech platforms rather than news organizations. And AI models rely on5256the legwork of journalists without fair compensation, going to the5257newsrooms and reporters who put in the work.5258    There has been a wave of consolidation in the media, ranging from5259broadcast affiliates to newspapers. And for many Americans, most of5260their news comes through social media algorithms, giving the owners of5261these platforms inescapable influence over what we see and read.52625263    Question 1. What policies should we consider to ensure the media5264and platforms serve individuals and support a democratic society,5265rather than undermine it?5266    Answer. Congress and the Supreme Court have long recognized that a5267democratic society cannot exist without a vibrant fourth estate where5268people have access to news from the ``widest possible dissemination of5269information from diverse and antagonistic sources.'' \1\ Accordingly,5270policies should not simply resist further consolidation or ensure the5271survival of existing news creators. Congress should affirmatively5272facilitate new, independent sources of news that can take full5273advantage of a multitude of platforms.5274---------------------------------------------------------------------------5275    \1\ Associated Press v. United States, 326 U.S. 1, 20 (1945). See5276also Turner Broadcasting Systems, Inc. v. FCC, 520 U.S. 180, 194 (1997)5277(``Federal policy, however, has long favored preserving a multiplicity5278of broadcast outlets regardless of whether the conduct that threatens5279it is motivated by anticompetitive animus or rises to the level of an5280antitrust violation'').5281---------------------------------------------------------------------------5282    First, however, it is important to distinguish between ``media''5283and ``platforms.'' Media--particularly legacy media such as newspapers5284and broadcast licensees, but also including digital news creators--are5285one-to-many outlets that engage in original reporting and distil news5286into a form understandable by the public. Platforms are a means of5287distributing content that do not themselves produce the content.5288Platforms can enable not only distribution of the work of journalists,5289but can enable robust discussion among the public--and preserve events5290that become the basis for reporters and serious journalists. At the5291same time, platforms can exercise unseen influence on how people see5292and encounter important news content. They can favor some sources of5293news or particular viewpoints without disclosing this bias. Platforms5294can provide an important income stream for reporting, but can also have5295negative impacts on income for journalists through demonitization and5296interfering with the relationship between advertisers and reporters (or5297the media that supports reporting, such as newspapers). This is5298particularly problematic with AI tools that provide a summary of news5299content developed by reporters and journalists.5300    While platforms have a First Amendment right to curate the content5301on their platforms. It is critical that Congress act to ensure that5302actual reporting by multiple and genuinely diverse and antagonistic5303outlets (that is to say, outlets in actual competition with each and5304therefore providing different perspectives, and with incentive to5305invest in reporting to ``scoop'' each other) is widespread and5306vigorous. As the Supreme Court has said, there is ``a governmental5307purpose of the highest order in ensuring public access to a5308multiplicity of information sources.'' \2\5309---------------------------------------------------------------------------5310    \2\ Turner Broadcasting Systems, Inc. v. FCC, 520 U.S. at 1905311(cleaned up).5312---------------------------------------------------------------------------5313    Congress should therefore adopt policies that promote competition5314between independent and diverse media sources, while at the same time5315ensuring adequate funding for reporting at the local, national and5316international level. At the same time, we must facilitate competition5317among platforms and make it easy for journalists to use a multitude of5318platforms to reach their audience. Finally, Congress must make sure5319that actual reporting remains economically sustainable.5320    For example, with regard to competition among media outlets,5321Congress should make clear that the FCC has no authority to raise the5322existing national audience reach of 39 percent, and should eliminate5323the ``UHF loophole'' that counts UHF stations at only 50 percent of5324their audience reach. Congress should also consider cross-ownership5325limitations that would prevent a single owner, or a small group of5326owners, from controlling both the productions and dissemination of news5327through broadcasters and then use control over platforms to favor its5328own viewpoint and disadvantage rivals.5329    Similarly, Congress should pass legislation designed to (a) prevent5330large platforms from monopolizing either journalists or advertisers;5331and (b) foster the development of new platforms with different content5332curation policies and competing algorithms. For example, requiring5333portability so that a news provider can take its audience to a rival5334platform (or easily replicate the same content on a rival platform),5335and prohibiting policies that demonitize news providers for using5336competitors, would allow news creators and distributors to maximize5337their revenue streams.53385339    Question 2. How can we level the playing field for independent5340local journalism against mammoth tech platforms?5341    Answer. Congress can take several affirmative steps, but should5342avoid other proposals that would actually harm the production of news5343from diverse and antagonistic sources. Public Knowledge has made5344several specific proposals to enhance antitrust enforcement which would5345directly address the ability of platforms to exercise market power in5346ways detrimental to local journalism (including making local5347journalists unduly dependent or support on referrals by dominant5348platforms). Additionally, Public knowledge has recommended using grant5349programs to support local journalism (with adequate safeguards to5350protect journalistic independence).\3\ To briefly summarize the most5351important elements of our proposals.5352---------------------------------------------------------------------------5353    \3\ See Lisa Macpherson and Morgan Wilsmann, ``A Policy Primer for5354Free Expression and Content Moderation, Part II: Empowering User5355Choice,'' Public Knowledge (December 9, 2024). https://5356publicknowledge.org/empowering-user-choice/53575358---------------------------------------------------------------------------5359   Local news grants53605361     Direct Federal support through local news grants5362            recognizes that local journalism provides public goods that5363            markets alone won't adequately support, particularly in5364            smaller communities where advertising revenue has never5365            been sufficient to sustain robust coverage. A superfund5366            model using fees collected from platforms benefiting from5367            news content, as a percentage of total users or some other5368            mechanism, offers a potential funding mechanism. By5369            collecting fees from digital platforms that profit from5370            news distribution and using those resources to support5371            local journalism broadly, this approach channels money back5372            toward news production without the government making5373            content-based decisions about which outlets deserve5374            support.53755376   Interoperability53775378     If journalists and news organizations could take their5379            audiences with them when platforms no longer serve them5380            well, platforms would face real competitive pressure to5381            treat publishers fairly. Currently, a journalist who builds5382            a following on a platform is locked in because leaving5383            means abandoning that audience. Interoperability and data5384            portability would let journalists maintain connections to5385            audiences across platforms, dramatically reducing platform5386            leverage.53875388   Enhance traditional antitrust remedies, creating multiple5389        platforms and diminishing the importance of any individual5390        platform to journalists generally.53915392     As we have seen in recent antitrust cases against Meta5393            and Google, traditional antitrust remedies are not5394            sufficient to address the market power of the largest5395            platform. Congress should consider legislation that would5396            make it easier for stakeholders to bring antitrust lawsuits5397            directly, and require structural remedies that would5398            address the market power of dominant platforms.53995400    Media Concentration. In today's media landscape, consumers are5401transfixed by their screens, whether by TV or mobile phone, to access a5402wide array of content, including news and entertainment. Yet ownership5403of the platforms that deliver that content is becoming increasingly5404concentrated in the hands of a few.5405    President Trump has said that the same family which already5406controls Paramount, CBS, and Skydance will lead the new ownership group5407for TikTok. Recent reports suggest that the same family is also seeking5408to acquire Warner Bros. Discovery--with the backing of the Trump5409administration.54105411    Question 1. How might these deals and other media mergers--which5412require the President's sign-off or his administration's approval--5413reshape our media and political landscape?5414    Answer. This raises tremendous concern over the ability of5415Americans to access genuinely diverse and antagonistic sources of5416information. The level of concentration spanning virtually every5417information and entertainment platform on its own would create a threat5418to democracy as the market power and financial interests of a single5419family places too much power to control the availability of news,5420access to real-time information, and influence through one of the5421largest and most popular social media platforms. The explicit5422machinations of the President of the United States, through secret5423dealings with regard to TikTok, and with public statements designed to5424manipulate the bidding process over Warner Brothers Discovery (coupled5425with the very real examples of how the President has used merger review5426to extract control mechanisms over news, advertising and other content)5427make explicit that the President expects his supporters at Skydance to5428actively serve as a combination of censor and propagandist for himself5429and his political allies.54305431    Question 2. How does increased media concentration affect the5432economics of local, independent journalism?5433    Answer. Acquisitions of local news outlets--whether local5434newspapers, local television stations, or local radio stations--have5435undermined the viability of local, independent journalism until it has5436become almost extinct. Whether hedge funds or publicly traded5437companies, these acquirers finance their unending acquisition sprees by5438accumulating ever greater debt. This, combined with the desire of these5439``absentee landlords'' to maximize profit regardless of the welfare of5440the local community, means layoffs of reporters and news crews in the5441name of ``synergies.'' Fewer reporters are expected to cover news for5442an increasing number of outlets. Local newsrooms of broadcast5443affiliates are reduced or eliminated entirely. These are often replaced5444with national ``must carry'' segments that reflect the political views5445of the large corporate owners rather than focusing on local issues.5446Actual reporting, which requires investment of resources and skilled5447journalists, is replaced with syndicated entertainment or opinion5448pieces that make no effort at journalistic practices.5449    In the worst cases, particularly with local newspapers, these giant5450corporate owners will simply shut down local newspapers rather than5451spend the money to maintain them. Broadcast news may be entirely5452replaced with cheap, syndicated entertainment programming. This has led5453to the creation of ``news deserts,'' where local communities lack any5454local reporting.54555456    Question 3. Is the administration following the requirement that5457Congress set out in the law to divest TikTok and its algorithm from5458Chinese control?5459    Answer. While it is clear that there have been violations of the5460law, the extent to which the President has violated the Protecting5461Americans from Foreign Adversary Controlled Applications Act (PAFACA)5462is impossible to fully determine because the President has neither5463briefed the public or Congress on the full details of the divestiture5464to ensure that the terms of the divestiture comply with the5465requirements of the Act. Even given what information is available,5466numerous parties have expressed concern that the leasing agreement with5467China for the algorithm, and ByteDance's continuing stake in the5468company, do not provide the necessary independence from Chinese control5469to satisfy the requirements of PAFACA, despite the President's5470certification.5471    There are two clear violations of PAFACA:54725473   PAFACA allotted just one 90 day extension to the president,5474        if the president could show considerable progress was being5475        made toward a divestiture deal. President Trump has issued5476        multiple executive orders claiming to delay enforcement for 75-5477        day periods without meeting the statutory requirements for the5478        90-day extension. The statute provides no authority for these5479        discretionary enforcement delays that aren't tied to5480        divestiture progress.54815482   While the divestiture deal is being discussed, TikTok5483        remained operating (apart from a short black-out stint), with5484        the executive failing to uphold the law by refusing to enforce5485        the provisions of PAFACA.54865487    Finally, while not an express violation of PAFACA, it was never the5488intent of Congress to allow the President to structure a deal behind5489closed doors to give control to his personal supporters.54905491    Platform Settlements. President Trump sued the two social media5492platforms testifying at this hearing because they removed him from5493their platforms after the January 6th insurrection.5494    Despite the fact that legal experts called these lawsuits5495meritless, both Meta and Google settled their lawsuits this year, once5496President Trump was back in office.54975498    Question 1. Mr. Feld, how do you think the lawsuits and settlements5499by Meta and Google affect what we see and hear on their platforms?5500    Answer. Meta and Google settled lawsuits concerning deplatforming5501Donald Trump after the President violated platform content policies--a5502perfectly legal action well within Meta and Google's rights. In fact,5503deciding what content and who can use their platforms is precisely5504first amendment protected. Nevertheless, through the act of settling,5505these platforms have made it clear that they are more than willing to5506amend their content policies to appease whichever party is in power to5507avoid regulatory scrutiny. We saw at the beginning of President Trump's5508second term in January 2025, when Meta announced it changed content5509moderation policies to ``promote free expression,'' by being more5510permissive of contentious topics around gender identity and5511immigration--opening the door wider to harassment of marginalized5512users. It also eliminated fact-checking, which President Trump and his5513allies believed is akin to censorship, despite fact-checking simply5514providing more speech to content. What results is a platform with more5515hate speech and toxic content, with fewer tools to address intentional5516falsehoods.55175518    FCC Censorship. There has been a lot of attention--and rightly so--5519on FCC Chairman Brendan Carr's threats to knock Jimmy Kimmel off the5520air. But Carr's censorship of Kimmel is just one in a series of attacks5521on the First Amendment. Although the FCC is supposed to be independent,5522under Brendan Carr, the agency has become the President's speech5523police.5524    Since taking over the FCC, Carr has reopened frivolous inquiries5525into CBS, NBC, and ABC about their coverage of the 2024 election;5526launched investigations into NPR and PBS stations; and publicly5527threatened broadcast TV and radio stations based on the content of5528their coverage of immigration issues.55295530    Question 1. Mr. Feld, is Chairman Carr's behavior consistent with5531the FCC's mandate to be independent of the White House?5532    Answer. No. Chairman Carr himself has characterized his role as5533being part of the President's agenda to remake the media by eliminating5534``DEI,'' and otherwise conforming media content to reflect the5535President's priorities. Chairman Carr has repeatedly echoed President5536Trump's social media posts on what media programming should be banned,5537and has echoed the President's complaints that public media and certain5538network programming are ``propaganda'' for the Democratic Party or5539unduly negative to President Trump.55405541    Question 2. Do you think that these investigations could be5542considered jawboning?5543    Answer. Absolutely. The combination of investigations with public5544statements by both the President and Chairman Carr that make it clear5545the investigations are retaliation for content and coverage they don't5546like is classic jawboning.55475548    Question 3. Does the First Amendment protect a broadcaster's5549ability to decide what they do or don't cover, and how they cover it?5550    Answer. The First Amendment directly protects the ability of a5551broadcaster to decide what they cover and how they cover it. See FCC v.5552League of Women Voters, 468 U.S. 364 (1984) (First Amendment prohibits5553Congress from preventing public broadcasters from editorializing). See5554also Arkansas Educational Television Commission v. Forbes, 523 U.S. 6665555(1998) (broadcaster may exclude independent candidate from televised5556Presidential candidate debate).5557    Because broadcast licenses are uniquely limited by the laws of5558physics, and those few broadcast licenses which can be granted without5559creating harmful interference are granted by the government, the5560government may impose certain obligations on broadcasters to allow5561others to respond to editorial positions expressed by broadcast5562licensees, or may require broadcast licensees to provide news and5563perspectives of importance to the local community. Compare Red Lion5564Broadcasting Co., Inc. v. FCC, 395 U.S. 367 (1969) (upholding Fairness5565Doctrine and right of reply) with Miami Herald Publishing Co. v.5566Tornillo, 418 U.S. 241 (1974) (striking down state right of reply5567statute when applied to newspaper). Even where the FCC is permitted to5568require a broadcaster to allow a right of response, however, it must do5569so in a content neutral and consistent manner, not in a manner designed5570to influence the editorial choices of the broadcaster. The Fairness5571Doctrine and public interest obligations applied to broadcasters are5572designed to create more speech, not to censor speech.55735574    Question 4. Does the First Amendment allow the FCC to revoke5575broadcast licenses solely because the Chairman, or the President of the5576United States, doesn't like what was broadcast?5577    Answer. Absolutely not. To revoke a license (or threaten to revoke5578a license) because the President of the United States dislikes what the5579broadcaster says is not merely a violation of the First Amendment (see5580League of Women Voters, supra), it violates Section 326 of5581Communications Act which prohibits ``any regulation or condition . . .5582which shall interfere with the right of free speech.'' \4\5583---------------------------------------------------------------------------5584    \4\ 47 U.S.C. Sec. 326.55855586    Attacks on the First Amendment: Congress created the Federal Trade5587Commission to protect consumers from unfair and deceptive practices and5588ensure a competitive free market. FTC Chairman Andrew Ferguson,5589however, has been using the FTC's authority to bully social media5590platforms into changing their content moderation decisions to favor5591conservative content and force companies to advertise on Elon Musk's X5592platform.5593    This is despite the fact that the Supreme Court has recognized that5594social media platforms have First Amendment rights to ``select and5595shape other parties' expression into their own curated speech5596products.''55975598    Question 1. Mr. Feld, what authority does the FTC have to dictate5599content moderation decisions to platforms?5600    Answer. None. The FTC Act, Clayton Act and Magnuson-Moss Warranty5601Act give the FTC the authority to police ``unfair or deceptive acts or5602practices'' and ``unfair methods of competition.'' This jurisdiction5603covers economic conduct, which does not include editorial judgements5604made by platforms. More specifically, platforms' content moderation5605decisions are expressive rights protected by the First Amendment.5606Unless the claims themselves are Section 5 ``deceptive'' in the5607commercial context (i.e., false advertising), the FTC has no lawful5608basis to intervene.56095610                                  [all]