“The Federal Government in the Age of Artificial Intelligence”

1[House Hearing, 119 Congress]2[From the U.S. Government Publishing Office]34                         THE FEDERAL GOVERNMENT5                 IN THE AGE OF ARTIFICIAL INTELLIGENCE67=======================================================================89                                HEARING1011                               BEFORE THE1213                         COMMITTEE ON OVERSIGHT14                         AND GOVERNMENT REFORM15                     U.S. HOUSE OF REPRESENTATIVES1617                    ONE HUNDRED NINETEENTH CONGRESS1819                             FIRST SESSION2021                               __________2223                              JUNE 5, 20252425                               __________2627                           Serial No. 119-312829                               __________3031Printed for the use of the Committee on Oversight and Government Reform3233[GRAPHIC NOT AVAILABLE IN TIFF FORMAT]3435                       Available on: govinfo.gov,36                         oversight.house.gov or37                             docs.house.gov3839                                __________4041                   U.S. GOVERNMENT PUBLISHING OFFICE4260-818 PDF                  WASHINGTON : 20254344-----------------------------------------------------------------------------------4546              COMMITTEE ON OVERSIGHT AND GOVERNMENT REFORM4748                    JAMES COMER, Kentucky, Chairman4950Jim Jordan, Ohio                     Vacant, Ranking Minority Member51Mike Turner, Ohio                    Eleanor Holmes Norton, District of52Paul Gosar, Arizona                      Columbia53Virginia Foxx, North Carolina        Stephen F. Lynch, Massachusetts54Glenn Grothman, Wisconsin            Raja Krishnamoorthi, Illinois55Michael Cloud, Texas                 Ro Khanna, California56Gary Palmer, Alabama                 Kweisi Mfume, Maryland57Clay Higgins, Louisiana              Shontel Brown, Ohio58Pete Sessions, Texas                 Melanie Stansbury, New Mexico59Andy Biggs, Arizona                  Robert Garcia, California60Nancy Mace, South Carolina           Maxwell Frost, Florida61Pat Fallon, Texas                    Summer Lee, Pennsylvania62Byron Donalds, Florida               Greg Casar, Texas63Scott Perry, Pennsylvania            Jasmine Crockett, Texas64William Timmons, South Carolina      Emily Randall, Washington65Tim Burchett, Tennessee              Suhas Subramanyam, Virginia66Marjorie Taylor Greene, Georgia      Yassamin Ansari, Arizona67Lauren Boebert, Colorado             Wesley Bell, Missouri68Anna Paulina Luna, Florida           Lateefah Simon, California69Nick Langworthy, New York            Dave Min, California70Eric Burlison, Missouri              Ayanna Pressley, Massachusetts71Eli Crane, Arizona                   Rashida Tlaib, Michigan72Brian Jack, Georgia73John McGuire, Virginia74Brandon Gill, Texas7576                                 ------77                       Mark Marin, Staff Director78                James Rust, Chief Counsel for Oversight79                     Mitch Benzine, General Counsel80            Duncan Wright, Senior Professional Staff Member81                      Josh Mathis, Senior Advisor82                      Lisa Mortier, Senior Advisor83                     Frederick Hill, Senior Advisor84      Mallory Cogar, Deputy Director of Operations and Chief Clerk8586                      Contact Number: 202-225-50748788                  Jamie Smith, Minority Staff Director89                      Contact Number: 202-225-50519091                                 ------9293                        C  O  N  T  E  N  T  S9495                              ----------9697                                                                   Page9899Hearing held on June 5, 2025.....................................     1100101                               WITNESSES102103                              ----------104Mr. Yll Bajraktari, President, Special Competitive Studies105  Project106    Oral Statement...............................................    12107108Mr. Bhavin Shah, Founder and Chief Executive Officer, Moveworks109    Oral Statement...............................................    13110111Ms. Linda Miller, Founder and Chief Growth Officer, TrackLight112    Oral Statement...............................................    15113114Mr. Adam Thierer, Senior Research Fellow, R Street Institute115    Oral Statement...............................................    17116117Mr. Bruce Schneier (Minority Witness), Fellow and Lecturer,118  Harvard Kennedy School119    Oral Statement...............................................    19120121Written opening statements and bios are available on the U.S.122  House of Representatives Document Repository at:123  docs.house.gov.124125                           INDEX OF DOCUMENTS126127                              ----------128129  * Statement for the Record, Cyera; submitted by Rep. Mace.130131  * Statement for the Record, Flashpoint; submitted by Rep. Mace.132133  * Article, Arizona Technology Council, ``Here's How AI is134  Flourishing in Arizona Tech Sector''; submitted by Rep. Biggs.135136  * Legislation, AZ HB2678; submitted by Rep. Biggs.137138  * Article, ABC News, ``Trump wants to send U.S. citizens to139  foreign prisons, legal experts say''; submitted by Rep.140  Crockett.141142  * Attendance Sheet, Full Committee Attendance Sheet for Hearing143  Suspended; submitted by Rep. Crockett.144145  * Image, COGR Attendance at Musk Subpoena Vote Suspension;146  submitted by Rep. Crockett.147148  * Statement for the Record, Chairman Comer; submitted by Rep.149  Mace.150151  * Report, ``Frontier Models are Capable of In-Context Scheming152  - Apollo Research''; submitted by Rep. Perry.153154  * Letter, re: Federal AI Moratorium - AITIC; submitted by Rep.155  Pressley.156157  * Report, ``Black Tech Agenda Advancing Equity and Reimagining158  Technology - Color of Change''; submitted by Rep. Pressley.159160  * E&C Vote Sheet, CRPT-119-IF00-Vote014-20250513; submitted by161  Rep. Trahan.162163The documents listed are available at: docs.house.gov.164                          ADDITIONAL DOCUMENTS165166                              ----------167168  * Questions for the Record: to Mr. Bajraktari; submitted by169  Rep. Gosar.170171  * Questions for the Record: to Mr. Bajraktari; submitted by172    Rep. Longworthy.173174  * Questions for the Record: to Ms. Miller; submitted by Rep.175  Gosar.176177  * Questions for the Record: to Mr. Schneier; submitted by Rep.178  Ansari.179180  * Questions for the Record: to Mr. Shah; submitted by Rep.181  Gosar.182183  * Questions for the Record: to Mr. Thierer; submitted by Rep.184  Gosar.185186  * Questions for the Record: to Mr. Thierer; submitted by Rep.187  Langworthy.188189These documents were submitted after the hearing, and may be190  available upon request.191192                         THE FEDERAL GOVERNMENT193                 IN THE AGE OF ARTIFICIAL INTELLIGENCE194195                              ----------196197                         Thursday, June 5, 2025198199                     U.S. House of Representatives200201              Committee on Oversight and Government Reform202203                                           Washington, D.C.204205    The Committee met, pursuant to notice, at 10:03 a.m., in206HVC-210, U.S. Capitol Visitor Center, Hon. Nancy Mace207presiding.208    Present: Representatives Mace, Jordan, Turner, Gosar, Foxx,209Grothman, Cloud, Palmer, Higgins, Sessions, Biggs, Fallon,210Donalds, Perry, Timmons, Burchett, Greene, Boebert, Luna,211Burlison, Crane, McGuire, Gill, Norton, Lynch, Krishnamoorthi,212Khanna, Mfume, Brown, Stansbury, Garcia, Frost, Lee, Casar,213Crockett, Randall, Subramanyam, Ansari, Bell, Simon, Min, and214Pressley.215    Also present: Representatives Trahan and Moskowitz.216    Ms. Mace. Good morning. The Committee on Oversight and217Government Reform will now come to order, and we welcome218everyone this morning.219    Without objection, the Chair may declare a recess at any220time.221    I recognize myself for the purpose of making an opening222statement.223    Good morning and thank you for joining us for this224discussion on the Federal Government and artificial225intelligence today. AI is no longer some futuristic idea. It is226here and it is already reshaping everything from health care227and national defense to finance and fraud prevention. The228Federal Government has a responsibility to harness this229technology to make government work faster and more efficiently230for the American people. Today, we will highlight the critical231uses of AI in the Federal Government, which create232efficiencies, improve services, and save taxpayers a whole lot233of money.234    The Department of Defense is using AI to improve decision-235making and protect our men and women in uniform. Federal236agencies are using AI to detect fraud before it happens by237using the technology to identify patterns of fraudulent238behavior and working proactively to prevent improper payments.239AI agents are beginning to help the administrative tasks,240freeing up Federal employees to focus on their mission instead241of paperwork. However, barriers remain and challenges must be242addressed so the government is to fully realize the benefits of243this transformative technology. For example, many outdated244legacy IT systems do not integrate with new technologies. A245cumbersome procurement process limits agency access to some of246the best new technologies, including those utilizing AI, and247poor data management leads to information silos and duplicative248work, which cost taxpayers money and prevents agencies from249realizing the full potential of AI.250    This Committee has continued to highlight these challenges251and will play a major role in addressing them, including by252continuing bipartisan legislative efforts from last Congress.253This includes the bipartisan Federal AI Governance and254Transparency Act led by Chairman Comer and then Ranking Member255Raskin, which laid out a vision for transparent and trustworthy256AI use in government. It also includes the AI Training257Extension Act, which I sponsored last Congress and reintroduced258today. This bill will equip Federal employees with the AI259literacy and skills necessary to leverage AI across the Federal260Government.261    I ask unanimous consent to enter into the record a262statement from the Chairman Comer highlighting the pivotal role263this Committee has to ensure agencies deploy AI responsibly and264the bipartisan efforts of the Committee on these issues.265Without objection.266    President Trump has rightly identified the importance of267U.S. dominance in AI and has taken action to remove unnecessary268barriers and unleash innovation. This is a refreshing reversal269from the heavy-handed regulation first approach to AI taken by270the Biden Administration. Last Congress, my Subcommittee on271Cybersecurity, Information Technology, and Government272Innovation held multiple hearings on the disastrous impacts of273overreaching policies, and I was relieved to see the former274President Biden's AI executive order rescinded on day one. This275was a critical first step, and since then, the White House has276directed agencies to prioritize the deployment of AI and277improve procurement processes to eliminate barriers to the278Federal Government's use of AI. I believe it is important for279this Committee to watch these developments closely and work to280ensure the executive branch has the tools and authorities to281deploy AI at scale and realize the full potential of this282incredible technology.283    I want to thank each of our witnesses this morning for284sharing their time with us today and look forward to hearing285from each of you. I also want to thank my colleagues from the286other side of the aisle who showed up in great number this287morning, too. I appreciate your interest in this topic. This288has been a bipartisan topic for this Committee. And with that,289I would like to recognize Ranking Member Lynch for his opening290statement.291    Mr. Lynch. Thank you very much. Good morning, Madam Chair,292and again, thank you for calling this hearing on the use of293artificial intelligence in our government.294    Optimizing the Federal Government's use of technology has295long been a bipartisan priority of this Committee. Indeed, as296you know, Madam Chair, our late colleague and friend, Gerry297Connolly, was a steadfast champion for using modern technology298to improve the delivery of government services, strengthen299oversight, and enhance public trust. We continue to fight his300fight on this and so many other critical challenges facing our301government. We cannot sit here, however, and have the302traditional bipartisan conversation about Federal IT303modernization without acknowledging the fact that the Trump304Administration, Elon Musk, and DOGE are leading technology305initiatives that threaten the privacy and security of all306Americans and undermine our government and the vital services307it provides to red states and blue states. Musk may say he has308stepped away from his role in the Federal Government, but his309recklessness will continue to have devastating consequences for310America for years, possibly decades to come.311    Today, we will work to better understand the disaster and312danger the Trump Administration has created by turning our313government over to his biggest campaign donor and people who314are more interested in self-enrichment than public service.315Since President Trump's inauguration, he has given Elon Musk316free rein to terrorize our civil servants and drive more than317275,000 Federal employees from their jobs serving the American318people. His DOGE team has done this so capriciously that they319have had to go back to people that they originally fired and320beg them to return to their jobs, serving our veterans,321maintaining our nuclear stockpile, and fighting bird flu. At322the Social Security Administration, whistleblowers have told us323that the Agency has lost so many critical information324technology and support staff, that very few of the remaining325workers even know how to handle errors or alerts in their326computer systems. This could mean missed checks and millions of327seniors and people with disabilities losing access to their328financial lifeline. While they purged the Federal Government of329expertise, Elon Musk and DOGE reportedly fed sensitive330government data, including Americans' private personal331information, into unvetted and unaccountable AI software. They332also reportedly deployed a Musk-owned generative AI chatbot333called Grok to consume government data and assist with334sensitive government decisions. Grok has even been deployed335onto systems at the Department of Homeland Security, despite336the fact that it has not been approved for use.337    Musk has been operating without any oversight whatsoever,338while posing a very real risk of violating security and privacy339laws, accessing confidential information about his competitors,340and using valuable Federal data bases to turbocharge his own AI341system, which would mean more money and more power for Mr.342Musk. The American people deserve answers. Musk and DOGE have343put the most personal data of every American at risk.344Cybersecurity experts like Mr. Schneier, who is here with us345today, are sounding the alarm that the holes DOGE has created346in our national cybersecurity system are gaping and pervasive.347And now, in this horrifying, insecure environment, the Trump348    Administration wants to centralize all of Americans' data349into one massive data base, which would allow full-spectrum350surveillance of American citizens, making it even easier for351bad actors to hack, and, of course, they have contracted with a352big, private tech company owned by one of President Trump's353billionaire supporters to do just that.354    Four months ago, Democrats moved to subpoena Elon Musk to355provide public testimony to this Committee. At that time, our356Republican colleagues complained that their constituents were357calling them, their offices, and showing up to town halls to358yell at them about Musk. Well, Americans' concerns have only359deepened since then. From his erratic purge of the Federal360workforce to his exploitation of sensitive taxpayer data, to361the cybersecurity nightmare he has created, and the horrifying362surveillance state we fear, Musk has put the American people363into a position where they demand answers from Elon Musk. I364believe, too, that my Republican colleagues should have had365some questions of their own for Mr. Musk.366    Recent allegations have left Americans wondering if he was367potentially under the influence of drugs while at the very368heart of the Trump Administration's most consequential and369sensitive decisions. It is worth investigating whether one of370President Trump's most influential advisors was under the371influence of heavy drugs while upending hundreds of thousands372of American lives, breaking government services beyond repair,373and handing out death sentences for hundreds of thousands of374innocent people around the world who relied on American foreign375aid and medical care for survival, most of whom were children.376    If this does not move my Republican colleagues, maybe they377just want to hear from Musk on why he called their bill to cut378Medicaid and give tax breaks to billionaires--this is Elon379Musk's words--``a disgusting abomination,'' and ``massive,380outrageous, pork filled congressional spending bill,'' or why381he spent the last 24 hours leveraging his social media to382``Kill the bill.'' Or maybe they want to know why Musk has said383that his next round of firings might be targeted at them,384Republican politicians who have not toed the line on Musk's385priorities.386    Regardless of your motivations, Democrats and Republicans387alike simply cannot allow Elon Musk to escape accountability388for his actions. He has dismantled our government, endangered389Americans, and weaponized public service for his own financial390gain. I ask my Republican colleagues, let us work together on391this. Let us support oversight that Democrats today in moving392that this Committee subpoena Elon Musk to answer for his393actions. All of us deserve answers on behalf of the American394people.395    So, Madam Chair, therefore, pursuant to Clause 2(k)(6) of396House Rule XI, I move that we subpoena, this Committee,397Republican and Democrat, subpoena Elon Musk to testify before398this Committee.399    Mr. Mfume. I second the motion.400    Ms. Mace. We will suspend for a moment. Thank you.401    Mr. Casar. Chairwoman, point of order. When are we going to402vote on subpoenaing Musk?403    Ms. Mace. We are doing a point of order right now with the404parliamentarian.405    Mr. Higgins. Madam Chair?406    Ms. Mace. Thank you. We are suspended.407    Mr. Krishnamoorthi. Madam Chair, may I be recognized?408    Ms. Mace. Yes, sir.409    Mr. Higgins. Madam Chair, I move to table the gentleman's410motion.411    Ms. Mace. OK. The motion is not debatable.412    As many as are in favor of tabling, signify by saying aye.413    [Chorus of ayes.]414    Ms. Mace. All those opposed, signify by saying no.415    [Chorus of noes.]416    Mr. Lynch. I ask for a recorded vote.417    Ms. Mace. In the opinion of the Chair, the ayes have it,418and the motion to table is agreed to.419    The Committee will now resume consideration----420    Mr. Lynch. I ask for a recorded vote.421    Mr. Casar. Interesting opinion.422    Mr. Krishnamoorthi. Madam Chair?423    Mr. Mfume. A recorded vote is requested.424    Mr. Krishnamoorthi. Parliamentary inquiry.425    Ms. Mace. OK. A recorded vote is ordered, and the clerk426will call the roll.427    OK. We are going to suspend to allow the clerk to get428ready.429    Mr. Krishnamoorthi. Parliamentary inquiry.430    Ms. Mace. Yes, sir.431    Mr. Krishnamoorthi. Madam Chair, thank you. Is it the432custom of the Committee to shut down debate on such an433important motion?434    Ms. Mace. This is not an inquiry. This is not an inquiry.435We are suspended.436    Mr. Krishnamoorthi. The gentleman from Massachusetts437mentioned that Mr. Musk was potentially under the influence of438ketamine----439    Ms. Mace. This is not debatable right now. I love you. It440is not debatable right now, and this is not a point of inquiry.441    Mr. Krishnamoorthi [continuing]. And other----442    Ms. Mace. But you look good in front of the photos. You443look great right now. You look good.444    Mr. Krishnamoorthi. So, is this the----445    Ms. Mace. You are not recognized right now. There is not a446point of inquiry. Thank you. We are suspended until the clerk447is ready to call the roll.448    Mr. Frost. Move to call the roll.449    Ms. Mace. We are suspended again. For the third time, we450are suspended until the clerk can call the roll. Thank you.451    [Pause.]452    Ms. Crockett. Madam Chair, point of inquiry.453    Ms. Mace. OK. Yes, ma'am.454    Ms. Crockett. Is the clerk here?455    Ms. Mace. The clerk is here and preparing to call the roll.456    Ms. Crockett. OK.457    Ms. Mace. Thank you.458    Ms. Crockett. Can you explain exactly what preparing459requires?460    Ms. Mace. She has to get paperwork. She has to get seated.461She has to do all that. Thank you. She sure does, yes.462    Ms. Stansbury. Or you guys have to find your Members who463did not attend the hearing. That is probably more likely.464    Mr. Krishnamoorthi. Madam Chair, inquiry, please.465    Ms. Mace. Yes, sir. Do you want more video with Elon behind466you? You already got a good clip.467    Mr. Krishnamoorthi. Could we have a unanimous consent468motion to structure debate for 10 minutes, each side gets 5469minutes to debate----470    Ms. Mace. We are getting ready to do a recorded vote.471    Mr. Krishnamoorthi [continuing]. Whether he was under the472influence of drugs?473    Ms. Mace. We cannot do that. We are not able to do that474right now. Let us do the recorded vote, then we can talk about475it after. Thank you.476    Ms. Crockett. Parliamentary inquiry.477    Ms. Mace. Yes, ma'am.478    Ms. Crockett. Do we have an assistant clerk that can move479maybe a little bit faster? We have a room full of people.480    Ms. Mace. This is not an inquiry. Thank you.481    Ms. Crockett. It was a question. Do we have an assistant482clerk?483    Ms. Mace. It is not a parliamentarian inquiry, so, no, we484are not having this debate.485    Mr. Mfume. But it is a point of order.486    Ms. Mace. It is not. A motion to table is not debatable.487Thank you.488    Ms. Crockett. We are not debating that. We are asking about489the clerk.490    Mr. Mfume. We were not asking for a point of order. Will491the parliamentarian rule on the point of order?492    Mr. Frost. Is the parliamentarian here?493    Ms. Mace. The gentleman must state his point of order.494    Mr. Mfume. It was the gentlelady's point of order that was495overruled because it was determined that it was not a point of496order when in fact it is. I think that she should be allowed497to----498    Ms. Mace. She stated it was an inquiry, not a point of499order, from my recollection, but if she would like to restate500it, that is fine.501    Mr. Mfume. Oh.502    Ms. Crockett. Madam Chair, point of order.503    Ms. Mace. Yes, ma'am.504    Ms. Crockett. Do we not have an assistant clerk who can505make sure that we can continue to do the business in an506efficient way in Oversight because it has historically never507taken this long for the clerk to call the roll.508    Ms. Mace. We are going through the process right now, and509we will have a recorded vote as soon as we are able to.510    Mr. Lynch. Just to restate the gentlelady's point,511Democrats are here. We think this is an important issue. We are512ready to vote.513    Ms. Mace. The gentleman must state a point of order.514    Mr. Lynch. And we welcome the opportunity to debate this.515This is a serious issue.516    Ms. Mace. Is there a point of order, Mr. Lynch?517    Mr. Krishnamoorthi. The gentleman has the Floor. Let him518finish.519    Ms. Mace. I am the Chairwoman. Thank you. Women are allowed520to speak on this Committee, and I ask for a point of order.521    Ms. Crockett. Point of inquiry. Madam Chair, point of522inquiry. I should, hopefully, get to speak?523    Ms. Mace. Yes, ma'am. Yes, you are. Yes, you are babe.524    Ms. Crockett. Is it possible for us to just do a roll call525without the clerk to determine what Members are here so that we526can determine if we all really have the votes?527    Ms. Mace. No, ma'am. Thank you.528    Ms. Stansbury. For those of you watching at home, the529Republicans are not in the room, so they are trying to find530their Members so that they can----531    Ms. Mace. You are not recognized, Ms. Stansbury. Thank you.532You can be quiet now.533    Ms. Stansbury. Neither are any of your Members because they534are not in the room. Thank you.535    Mr. Lynch. Madam Chair, on a point of order, Democrats have536a clerk that is available to record a vote.537    Ms. Mace. OK. Mr. Lynch, the Chair will not entertain538points of order regarding the propriety or expediency of a539proposed course of action. This is my final ruling on such540points of order. Thank you.541    Mr. Lynch. Madam Chair, a point of inquiry then.542    Ms. Mace. Yes, sir.543    Mr. Lynch. Would it not be permissible to have the544Democratic clerk record the vote and to begin it now?545    Ms. Mace. No, sir, it is not. Thank you.546    Ms. Crockett. Madam Chair?547    Ms. Mace. Yes, ma'am.548    Ms. Crockett. Is the only reason we cannot take a vote is549because you know you are about to lose it?550    Ms. Mace. Is there a point of inquiry here, because I did551not hear one.552    Ms. Crockett. That was my inquiry. Is the reason we----553    Ms. Mace. That is not a valid inquiry. Thank you.554    Ms. Crockett. OK.555    Mr. Min. I have a point of inquiry, Madam Chair.556    Ms. Mace. Yes, sir.557    Mr. Min. I was just wondering, being new to this Committee,558what is the basis or rule on which you are not allowing the559Democratic clerk to start the vote?560    Ms. Mace. We have our own clerk. That is why.561    Mr. Min. Is that based on the rule or?562    Mr. Crane. It is based on the fact that you guys lost the563election because of your crazy policies. And you guys do not564control the chamber.565    Mr. Min. Is that in the rules? Madam Chair, I just wanted566to follow-up, would appreciate an answer to the question. We567have witnesses who have flown in from around the country.568    Ms. Mace. This is Committee practice. Thank you for your569inquiry, and we will start the vote momentarily. Thank you.570    Mr. Lynch. Madam Chair, just a point of inquiry.571    Ms. Mace. Yes, sir.572    Mr. Lynch. It has now been 10 minutes, and this motion is573simply to bring Mr. Musk forward so that we can ask him574questions on behalf of the American people. I think the575Democratic position has been very reasonable, very deliberate.576You have got a lot----577    Ms. Mace. The motion is to table, and there is no debate.578Thank you, Mr. Lynch.579    Mr. Lynch. Again, Madam Chair, Democrats are all here, and580we are ready to vote. It is now 11 minutes.581    Ms. Mace. You are not recognized, but you do look good for582your clips later, for social media, with Elon and all the words583behind you. It looks good.584    Mr. Lynch. Appreciate it.585    Ms. Mace. The blue is good with the red color.586    Mr. Lynch. I appreciate that. Thank you. I really587appreciate that. Thank you.588    Mr. Fallon. Madam Chair, parliamentary inquiry. In a589minute, will it be another minute?590    Mr. Burchett. Probably 12, right?591    Ms. Mace. Could be.592    Mr. Fallon. It is debatable, depending on the math, I593suppose.594    Ms. Crockett. Madam Chair. Madam Chair, I have a unanimous595consent.596    Ms. Mace. We are in the middle of getting ready to vote, so597we cannot do that right now. We can do that after.598    Ms. Crockett. Is that the rule because the clerk has not599been ready for the last 15 minutes, almost.600    Ms. Mace. You are not recognized. Thank you.601    Ms. Crockett. So, we are not doing unanimous consent.602    Ms. Mace. Not in the middle of a motion to table and a603recorded vote we are preparing to take.604    Ms. Crockett. Is the clerk seated?605    Ms. Mace. You are not recognized right now. Thank you.606    Ms. Crockett. OK. Since we are sitting here in silence for607the----608    Ms. Mace. You are not recognized, and there is no point of609inquiry. You are not recognized.610    Ms. Crockett. OK. I will do a point of inquiry.611    Ms. Mace. OK.612    Ms. Crockett. I am seeking an inquiry to determine whether613or not it is OK to put it into the record that at the time that614this motion was called, only Mr. Gosar, Mr. Grothman, Mr.615Higgins, Mr. Fallon, and Mr. Crane, along with yourself, were616the only Republicans that were actually in the room.617    Ms. Mace. This is not a real point of inquiry. You are not618recognized on it either. Thank you.619    Ms. Pressley. Madam Chair, I have a legitimate point of620inquiry. Madam Chair.621    Ms. Mace. Is it for real, this time?622    Ms. Pressley. Oh, no, it is very legitimate, very for real.623How much taxpayer money are Republicans wasting while delaying624this vote to subpoena Elon Musk?625    Ms. Mace. That is not a legitimate point of inquiry.626    Ms. Pressley. It is very legitimate to the American people.627    Ms. Mace. It is not. Thank you. It is not. You are not628recognized.629    Ms. Pressley. Very legitimate to the American people.630    Ms. Mace. It is not, and you are not recognized on it.631Thank you.632    Mr. Fallon. Madam Chair, parliamentary inquiry.633    Ms. Mace. Yes, sir.634    Mr. Fallon. How much taxpayer money was wasted when some of635the Democrats did not show up to work for almost 2 years during636COVID?637    Ms. Mace. Also, not a point of inquiry. I love you, Mr.638Fallon.639    Mr. Fallon. Particularly one from Boston.640    Ms. Mace. Also, not a point of inquiry. Thank you.641    Ms. Pressley. Who are you talking about? I am sorry, are642you engaging in personalities right now? Who are you talking643about? Who are you talking about?644    Mr. Fallon. I did not say anybody's name.645    Ms. Pressley. Who are you talking about?646    Ms. Mace. This is not time for debate right now. Thank you.647    Ms. Pressley. Well, please.648    Ms. Mace. I am being respectful of everybody on both sides649of the aisle. Let us be respectful of one another, please.650    Mr. Fallon. I am from Pittsville, and I was here.651    Ms. Mace. This is not debatable. We are not having a debate652on this. We are getting ready to take a recorded vote.653Everybody can sit down and turn their microphones off unless654you have a point of inquiry, an appropriate point of inquiry.655Thank you.656    Ms. Crockett. Madam Chair, point of inquiry.657    Ms. Mace. Yes, ma'am.658    Ms. Crockett. You said that we were about to take the vote.659Oh, the clerk.660    Ms. Mace. Yes, we are about to take the vote.661    Ms. Crockett. Never mind.662    Ms. Mace. Thank you.663    Ms. Crockett. Because I was about to say, your Members664finally showed up. I mean, let us go.665    Ms. Mace. Thank you, ma'am. All right. A recorded vote is666ordered and the clerk will call the roll.667    The Clerk. Mr. Jordan?668    Mr. Jordan. Yes.669    The Clerk. Mr. Jordan votes yes.670    Mr. Turner?671    [No response.]672    The Clerk. Mr. Gosar?673    [No response.]674    The Clerk. Ms. Foxx?675    Ms. Foxx. Yes.676    The Clerk. Ms. Foxx votes yes.677    Mr. Grothman?678    Mr. Grothman. Yes.679    The Clerk. Mr. Grothman votes yes.680    Mr. Cloud?681    Mr. Cloud. Yes.682    The Clerk. Mr. Cloud votes yes.683    Mr. Palmer?684    Mr. Palmer. Yes.685    The Clerk. Mr. Palmer votes yes.686    Mr. Higgins?687    Mr. Higgins. Yes.688    The Clerk. Mr. Higgins votes yes.689    Mr. Sessions?690    Mr. Sessions. Aye.691    The Clerk. Mr. Sessions votes aye.692    Mr. Biggs?693    [No response.]694    The Clerk. Mr. Fallon?695    Mr. Fallon. Aye.696    The Clerk. Mr. Fallon votes aye.697    Mr. Donalds?698    Mr. Donalds. Yes.699    The Clerk. Mr. Donalds votes yes.700    Mr. Perry?701    Mr. Perry. Aye.702    The Clerk. Mr. Perry votes aye.703    Mr. Timmons?704    Mr. Timmons. Yes.705    The Clerk. Mr. Timmons votes yes.706    Mr. Burchett?707    Mr. Burchett. Yes.708    The Clerk. Mr. Burchett votes yes.709    Ms. Greene?710    Ms. Greene. Yes.711    The Clerk. Ms. Greene votes yes.712    Ms. Boebert?713    [No response.]714    The Clerk. Mrs. Luna?715    Mrs. Luna. Aye.716    The Clerk. Mrs. Luna votes aye.717    Mr. Langworthy?718    [No response.]719    The Clerk. Mr. Burlison?720    Mr. Burlison. Yes.721    The Clerk. Mr. Burlison votes yes.722    Mr. Crane?723    Mr. Crane. Yes.724    The Clerk. Mr. Crane votes yes.725    Mr. Jack?726    [No response.]727    The Clerk. Mr. McGuire?728    Mr. McGuire. Yes.729    The Clerk. Mr. McGuire votes yes.730    Mr. Gill?731    Mr. Gill. Yes.732    The Clerk. Mr. Gill votes yes.733    Mr. Lynch?734    Mr. Lynch. No.735    The Clerk. Mr. Lynch votes no.736    Ms. Norton?737    Ms. Norton. No.738    The Clerk. Ms. Norton votes no.739    Mr. Krishnamoorthi?740    Mr. Krishnamoorthi. No.741    The Clerk. Mr. Krishnamoorthi votes no.742    Mr. Khanna?743    Mr. Khanna. No.744    The Clerk. Mr. Khanna votes no.745    Mr. Mfume?746    Mr. Mfume. No.747    The Clerk. Mr. Mfume votes no.748    Ms. Brown?749    Ms. Brown. No.750    The Clerk. Ms. Brown votes no.751    Ms. Stansbury?752    Ms. Stansbury. No.753    The Clerk. Ms. Stansbury votes no.754    Mr. Garcia?755    Mr. Garcia. No.756    The Clerk. Mr. Garcia votes no.757    Mr. Frost?758    Mr. Frost. No.759    The Clerk. Mr. Frost votes no.760    Ms. Lee?761    Ms. Lee. No.762    The Clerk. Ms. Lee votes no.763    Mr. Casar?764    Mr. Casar. No.765    The Clerk. Mr. Casar votes no.766    Ms. Crockett?767    Ms. Crockett. No.768    The Clerk. Ms. Crockett votes no.769    Ms. Randall?770    Ms. Randall. No.771    The Clerk. Ms. Randall votes no.772    Mr. Subramanyam?773    Mr. Subramanyam. No.774    The Clerk. Mr. Subramanyam votes no.775    Ms. Ansari?776    Ms. Ansari. No.777    The Clerk. Ms. Ansari votes no.778    Mr. Bell?779    Mr. Bell. No.780    The Clerk. Mr. Bell votes no.781    Ms. Simon?782    Ms. Simon. No.783    The Clerk. Ms. Simon votes no.784    Mr. Min?785    Mr. Min. No.786    The Clerk. Mr. Min votes no.787    Ms. Pressley?788    Ms. Pressley. No. No.789    The Clerk. Ms. Pressley votes no.790    Ms. Tlaib?791    [No response.]792    The Clerk. Madam Chairwoman?793    Ms. Mace. Aye.794    The Clerk. Madam Chairwoman votes aye.795    Ms. Mace. And how is Mr. Biggs recorded?796    The Clerk. Mr. Biggs is not recorded.797    Mr. Biggs. Aye.798    The Clerk. Mr. Biggs votes aye.799    Ms. Mace. And how is Ms. Boebert recorded?800    The Clerk. Ms. Boebert is not recorded.801    Ms. Boebert. Aye.802    The Clerk. Ms. Boebert votes aye.803    Mr. Frost. Madam Chair, parliamentary inquiry.804    Ms. Mace. The clerk will report the tally.805    The Clerk. Madam Chairwoman, on this vote, the ayes are 21.806The nays are 19.807    Ms. Mace. The ayes have it, and the motion to table is808agreed to.809    The Committee will now resume.810    OK. I now request unanimous consent that Representative811Moskowitz of Florida and Trahan of Massachusetts will be waived812on to today's hearing for the purpose of asking questions, and813without objection, so ordered.814    I am pleased to introduce our witnesses for today's815hearing. Our first witness today is Mr. Yll Bajraktari, the816President of the Special Competitive Studies Project. Our817second witness is Mr. Bhavin Shah, founder and chief executive818of the AI company, Moveworks. Our third witness is Ms. Linda819Miller, former Deputy Executive Director of the Pandemic820Response Accountability Committee and founder of TrackLight.821Our fourth witness is Mr. Adam Thierer, Senior Research Fellow822at R Street Institute, and our fifth witness today is Mr. Bruce823Schneier from the Harvard Kennedy School. We welcome everyone,824and we are pleased to have you here this morning.825    So, pursuant to Committee Rule 9(g), the witnesses will826please stand and raise your right hands.827    Do you solemnly swear or affirm the testimony you are about828to give is the truth, the whole truth, and nothing but the829truth, so help you God?830    [A chorus of ayes.]831    Ms. Mace. Let the record show the witnesses all answered in832the affirmative.833    We appreciate all of you being here today and look forward834to your testimony. Let me remind the witnesses that we have835read your written statements, and they will appear in full on836the hearing record. Please limit your oral statements to 5837minutes. As a reminder, please press the button on the838microphone in front of you so that it is turned on and the839Members up here can hear you. When you begin to speak, the840light in front of you will turn green. After 4 minutes, the841light will turn yellow. When the red light comes on, your 5842minutes has expired, and we would ask that you please wrap it843up.844    So, I now recognize Mr. Bajraktari to please begin your845opening statement.846847                      STATEMENT OF YLL BAJRAKTARI848849                               PRESIDENT850851                  SPECIAL COMPETITIVE STUDIES PROJECT852853    Mr. Bajraktari. Chairwoman Mace, Ranking Member Lynch, and854Members of the Committee, thank you for the opportunity to855testify here today. I am here to make one clear point. While856developing advanced AI is a significant moment, the true857measure of a Nation's future advantage will not be who has the858best models, but who has built the essential digital859infrastructure and fosters the deepest adoption of AI across860all of the society. To understand the stakes, imagine if at the861beginning of the last century, our Nation had failed to embrace862electricity and yielded that transformative advantage to our863competitors. The consequences would have been devastating.864    Today, we face a similar moment. The strategic competition865in AI with People's Republic of China is intensifying. Since8662017, Beijing has launched an ambitious campaign to achieve867technological supremacy as the key to its global dominance.868China is executing this strategy by dedicating vast resources869to integrate AI across its entire economy, society, and870military. For example, they recently announced $47.5 billion871semiconductor fund, which is designed to eliminate reliance on872Western technology. They are on track to surpass the U.S. in873raw R&D spending, having already tripled their investments874since Xi Jinping came to power in 2012 to $500 billion in 2024.875    This is all built on dominant infrastructure, including876more than 60 percent of the world's 5G base stations. This877entire effort is backed by state-sponsored cyber espionage and878intellectual property theft designed to acquire and weaponize879advanced technology. While the United States still leads AI in880research, our primary disadvantage is the slow adoption of881these breakthroughs. We are hampered by bureaucratic inertia,882outdated IT infrastructure, and a lack of AI literacy in the883workforce. Overcoming these barriers is the key to winning the884global tech competition. To secure our leadership, United885States must take decisive action in the following areas.886    First, we need to establish a high-level technology887competitiveness council at the White House modeled on the888approach we took for the space race to coordinate a unified889national strategy and drive rapid AI adoption across the890government. Second, we need to significantly increase our891investment in AI infrastructure by doubling non-defense AI R&D892spending to $32 billion over time and codifying the national AI893research and resource. Third, we need to launch a comprehensive894national talent strategy to make the U.S. the magnet for AI895experts. This means eliminating green card caps for the top896STEM talent and promoting AI literacy throughout our897educational system. Fourth, we must overhaul the Federal898procurement process to rapidly integrate cutting-edge AI from899our most innovative companies. Last, we must strengthen our900global alliances in AI, 5G, and cybersecurity to ensure the901resilience and security for us and our allies.902    The bottom line is this: this is not just a competition of903invention, but of adoption. It is about who can set the rules904for the future. We have faced moments like this before, from905the space race to the digital revolution. Our Nation's greatest906strength has always been our ability to unite our vision with907decisive action. By embracing this challenge together across908government, industry, and academia, we will not only secure our909leadership, but build a more prosperous and secure future for910all Americans. Thank you, and I look forward to your questions.911    Ms. Mace. Thank you. I will now recognize Mr. Shah for his912opening statement.913914                        STATEMENT OF BHAVIN SHAH915916                  FOUNDER AND CHIEF EXECUTIVE OFFICER917918                               MOVEWORKS919920    Mr. Shah. Chairwoman Mace, Ranking Member Lynch, and921distinguished Members of the Committee, thank you for the922opportunity to testify on government use and procurement of AI.923I am Bhavin Shah, CEO and founder of Moveworks. Moveworks is an924enterprise AI company that has spent 9 years successfully925helping some of the world's largest organizations to use AI to926transform their operations. Today, we face a moment of national927importance. The question before us is not whether the Federal928Government should adopt AI. It is whether we will lead or929follow. Before we talk about the procurement hurdles that we930have faced, let me explain to you what Moveworks does.931    Moveworks transforms how organizations support their932employees. We do this through an enterprise AI assistant, and933employees interact conversationally with the assistant in a934chat-based interface. They can get IT support, ask H.R.935questions, and find company information. But the assistant is936not just for looking up information. It can also be used to937take actions, such as scheduling time off or obtaining software938licenses. Behind the scenes, our platform integrates with all939of an organization's technology systems.940    Because the AI assistant is a single unified tool,941employees no longer have to log in and out of multiple systems942to get work done. Organizations can extend the system to943simplify and accelerate their most complex business processes,944all done safely and securely while respecting permissions. We945do this work by combining top-tier language models, like946OpenAI's GPT-4, Meta Llama 3, as well as other models from947Google's DeepMind, Microsoft, NVIDIA, just to name a few. We948put these together with our own enterprise-specific models, all949governed by enterprise-grade guardrails, including fact-950checking, citations, and security filters. The result is AI951that understands work context. It handles complex business952processes safely and effectively and follows industry-leading953data protection and privacy approaches.954    Today, Moveworks serves 5 million employees across 350955organizations, including major companies like Broadcom,956Marriott, and CVS Health. We are also used by a number of local957governments, such as Scottsdale, Ventura, and Durham County. In958addition, government contractors, like Leidos and Northrop959Grumman, use Moveworks to make their businesses more efficient.960We have seen firsthand how AI drives measurable results.961Honeywell reduced IT help desk requests by 80 percent, Broadcom962scaled from 10,000 to 50,000 employees without increasing963service staff, and the city of Glendale achieved a 514 percent964return on its AI investment.965    Our experience positions us uniquely to understand the966Federal Government's complex requirements. The Federal967Government has an unprecedent opportunity. You can leverage AI968for efficiency, you can achieve cost savings, you can improve969employee experience, and, most importantly, you can therefore970improve citizen services. We have engaged with agencies across971the government and consistently see enthusiasm for AI adoption,972but that enthusiasm has not translated into government973procuring AI products like Moveworks.974    We have encountered three primary barriers. The first is975the FedRAMP process. Moveworks invested 3-and-a-half years and976over $8-and-a-half million to achieve FedRAMP-ready status.977This is a prohibited barrier for smaller AI innovators, the978companies that often develop the most cutting-edge solutions.979Second, complex reseller networks. These force innovative980companies to work through intermediaries rather than directly981with agencies, adding cost without adding value. Third, multi-982year contracting vehicles. These favor large incumbents over983agile AI companies. These create delays measured in years.984Meanwhile, AI technology advances every month.985    American AI leadership requires action, not just intention.986The private sector has demonstrated that AI adoption can both987be rapid and secure when procurement processes match the pace988of innovation. I recommend Congress consider immediate steps to989improve government use and procurement of AI. Embed AI990education and adoption goals in agency leadership performance991plans. Create government developer programs for startups to992learn Federal systems. Scale AI proof-of-concept programs for993rapid pilots to all agencies, similar to the Investment994Horizon's pilot program developed by the Office of the CIO995within the Navy. And finally, establish agency innovator996programs that allow smaller agencies to serve as test beds for997AI implementations. Longer term, we must modernize acquisition998policies. We must accelerate procurement timelines. The Federal999Government employs the most important workforce in the world,1000the workforce that operates our democracy and defends our1001Nation. These public servants deserve the same AI-powered tools1002that are transforming the private sector's productivity and1003effectiveness.1004    We have a choice. We can modernize our procurement process1005to embrace AI innovation or we can watch other Nations take the1006lead in government AI adoption while our agencies struggle with1007outdated processes and missed opportunities. I urge this1008Committee to champion procurement reform. We need to maintain1009our competitive edge while serving citizens more effectively.1010    Thank you, and I welcome your questions and discussions on1011these critical topics.1012    Ms. Mace. Thank you. I will now recognize Ms. Miller for1013her opening statement.10141015                       STATEMENT OF LINDA MILLER10161017                    FOUNDER AND CHIEF GROWTH OFFICER10181019                               TRACKLIGHT10201021    Mr. Miller. Thank you, Chairwoman, Ranking Member, and1022Members of the Committee. My name is Linda Miller. I bring a1023unique perspective today because I spent 10 years at one of the1024world's most admired agencies, the GAO, as well as I am also1025the co-founder of an AI technology startup. So, I am coming1026here with both the perspective of a long history in the1027oversight community, as well as some of the challenges that1028have already been raised about trying to bring innovative1029technology to the U.S. Government.1030    As DOGE is seeing firsthand, it is hard, if not impossible,1031to try and impose private sector innovation on the Federal1032Government. The data challenges alone are formidable. Agencies1033collect data poorly if they collect it at all. Programs collect1034their own data in silos. They do not share it even amongst1035themselves, much less with other agencies. In many cases,1036agencies do not even know who they are paying. They send money1037to states in the form of grants, and from there, the money1038trail is opaque, making payment integrity almost impossible. At1039TrackLight, to help illuminate who one Federal agency was1040paying, we spoke to a company who FOIAs data on invoices to1041state agencies and sells it. For our Federal client, we were1042going to have to buy the data to tell them who they were1043paying. The same issue exists at the state level. GAO reported1044on a grant program that it administered by one state in which a1045company applied for the same grant 13 times in neighboring1046counties and got it because the states did not share that data.1047    The size and complexity of the administrative state and the1048volume of regulations agencies must comply with make agility1049and innovation next to impossible. My fellow panelists have1050made some very smart suggestions about turbocharging the rip1051and replace efforts of legacy IT systems and removing1052procurement barriers, but we have to be realistic about how1053capable government is to enact these rapid changes no matter1054how badly they are needed. I could spend my time this morning1055telling you about how we urgently need to adopt AI across1056government to remain relevant, achieve cost savings, and1057improve service delivery as well as prevent fraud, waste, and1058abuse, but I think you already know that. So, I would like to1059talk about what proactively can be done to move the needle.1060    GAO has reported that there are almost 25 existing AI1061requirements in Federal law and guidance today. GSA has1062established an AI center of excellence which was required by1063law. Agencies have inventoried AI use cases. They have put in1064place responsible AI officials, but the government continues to1065show anemic AI adoption overall. My fear is we have created yet1066another bureaucratic morass, hamstringing agencies with1067compliance work and only added the layers of sludge, further1068slowing down AI adoption. There has been progress. Recent EOs1069have sought to improve data sharing, which will improve how1070agencies better make use of data to use AI, and GSA's FedRAMP1071overhaul is a step in the right direction. I completely agree1072with Mr. Shah. He spent $8.5 million trying to get FedRAMP1073certified and successfully getting FedRAMP certified. My small1074technology company is going to have to put out millions of1075dollars to do the same.1076    But AI adoption is the perfect candidate for what is called1077a regulatory sandbox, which Congress can mandate to speed AI1078projects to implementation in controlled settings, suspending1079these layers of bureaucracy. Creative procurement solutions1080must be explored and carefully monitored, allowing lessons to1081be learned and shared across agencies. And outdated privacy1082laws, privacy laws that were designed for a world that existed1083long before the internet, much less artificial intelligence,1084has to be revised. I cannot state this more clearly: the bad1085guys have all our data. We cannot protect Americans' privacy in10862025. Today, privacy protection is going to have to involve1087advanced technology, like anonymization of sensitive data, and1088the use of other types of identifiers like behaviors and1089geolocation.1090    I would be remiss not to mention the risks, which are real.1091Lazy, careless use of AI does not just imperil the goals of the1092AI project, it undermines the trust in the technology itself.1093Hallucinations are real. Subject matter expertise is vital.1094Responsible AI is not just a buzzword. Humans in the loop means1095humans who are trained, knowledgeable, and paying close1096attention. This is not a set it and forget it tool. Now is the1097time to be bold. We cannot change how government operates.1098Wholesale changes to legacy IT systems and the Federal1099acquisition system will take years, but we can create1100innovative laboratories where AI projects can operate in proof-1101of-concept, regulatory sandboxes in carefully controlled1102environments to start to show the art of the possible.1103    I welcome your questions. Thank you.1104    Ms. Mace. Thank you. I will now recognize Mr. Thierer for1105his opening statement.11061107                       STATEMENT OF ADAM THIERER11081109                         SENIOR RESEARCH FELLOW11101111                           R STREET INSTITUTE11121113    Mr. Thierer. Chairwoman Mace, Ranking Member Lynch, and1114Members of the Committee, thank you for your invitation to1115participate in this important hearing. My name is Adam Thierer.1116I am a senior fellow at the R Street Institute where I cover1117emerging technology policy. My message here today focuses on1118three key points. First, there are meaningful benefits to1119governmental use of artificial intelligence technologies.1120Second, Congress and the Trump Administration need to take1121steps to unlock those benefits by accelerating the1122modernization of government systems and policies. Third, we1123must appreciate the connection between broader AI regulation1124and the benefits the government itself can accrue from these1125systems.1126    While the Federal Government is already integrating AI into1127systems and processes, progress has been slow, and there needs1128to be more urgency to tap into the many benefits that AI can1129offer. The bipartisan House Task Force on Artificial1130Intelligence report from last December noted that, ``Each year,1131the Federal Government spends over $100 billion on information1132technology and cybersecurity. Approximately 80 percent of this1133spending goes to operating existing legacy systems that are1134typically outdated and underpinned by archaic software and1135hardware components.'' AI can help address this persistent1136problem.1137    Last year, the U.S. Chamber of Commerce released a study on1138how improved IT processes in eight already helped to1139``modernize infrastructure, enhance citizen services, improve1140national security, and foster innovation, and unlock1141significant long-term cost savings and efficiencies along the1142way.'' Doing more with AI could generate even broader benefits.1143Accenture estimates that by tapping AI systems, Federal1144agencies could unleash ``a productivity windfall for the U.S.1145Government worth up to $532 billion annually by 2028.''1146    Just earlier this week, the U.K. Government reported that a1147new trial of AI tools revealed how civil servants who used them1148freed up 2 weeks a year in working time, and 82 percent of them1149said they wanted to continue to use AI tools to boost their1150productivity. AI tools can be particularly helpful in1151simplifying paperwork and procedural hassles. AI has already1152been used to improve public records management at many1153different agencies, including document declassification and1154Freedom of Information Act requests. AI can also help1155streamline specific regulations and make them more cost1156effective to ensure greater value to taxpayers. AI is already1157helping governments better administer various public services,1158including infrastructure management, public safety efforts,1159healthcare delivery, environmental monitoring, emergency1160responses, workforce development, and various other citizen1161services.1162    As Congress and the executive branch look to continue to1163clear a path to efficient government AI uptake, the focus1164should continue to be on addressing barriers to implementation1165through five main priorities. First, modernize Federal1166acquisition policies to boost AI adoption, especially by1167encouraging easier acquisition of sophisticated but cost-1168effective commercial systems. Second, streamline paperwork and1169use requirements associated with Federal AI contracts and1170deployments, especially for small businesses and open-source1171providers. Third, ensure government data sets are AI ready and1172interoperable such that the public, researchers, and other1173organizations can more easily benefit from them. Fourth,1174enhance in-house government AI talent and improve technical1175literacy at all levels. And fifth, boost trust and security in1176government technology systems.1177    Congress has already passed several laws to facilitate some1178of these goals, and bipartisan efforts by both the Biden and1179Trump Administrations have helped expand those efforts,1180including through chief AI officers in Federal agencies. The1181OMB recently released very important guidance requiring1182agencies to ``adopt a forward-leaning and pro-innovation1183approach that takes advantage of this technology to help shape1184the future of government operations,'' and ``harness solutions1185that bring the very best value to taxpayers.'' This Committee1186and others should assist these efforts by helping the1187Administration ensure that agencies are AI ready for the1188future. That means closer oversight of agency modernization and1189digitization efforts plus the necessary funding to get the job1190done.1191    Finally, Congress should recognize there is an important1192connection between broader AI policy and the benefits1193governments can gain from AI systems. Proposals to regulate AI1194systems are proliferating rapidly with over 1,000 AI-related1195bills already introduced just 5 months into 2025. If these1196mandates expand, they will significantly raise the cost of1197deploying advanced AI systems because complicated, confusing1198compliance regimes would hamstring developers, especially1199smaller ones. A confusing patchwork of state and local red tape1200will not only undermine the market for commercial systems, but1201it will also undermine the government's ability to choose from1202a diverse array of competitive, cost-effective options.1203    One recent study from CCIA reported the Federal preemption1204of state-level AI regulation would benefit the Federal balance1205sheet by up to $269 billion over the next decade by lowering1206procurement costs for the Federal Government. Congress needs to1207protect the interstate marketplace and ensure a robustly1208innovative and competitive AI ecosystem can develop. Some1209degree of preemption is needed to achieve this goal and ensure1210a diverse array of cutting-edge AI solutions are available to1211government agencies and to the public as consumers of both1212private and public services.1213    Thank you so much for inviting me here today. I look1214forward to your questions.1215    Ms. Mace. Thank you, and, Mr. Schneier, you are now1216recognized for your opening statement.12171218                      STATEMENT OF BRUCE SCHNEIER12191220                          FELLOW AND LECTURER12211222                         HARVARD KENNEDY SCHOOL12231224    Mr. Schneier. Thank you, and thanks for inviting me. So, I1225am a technologist. I have been working in cryptography and1226cybersecurity since the early 1990s. I write, I teach, I1227consult, I start companies. I thought a lot about the role of1228AI in democratic government. I actually wrote a book on this1229topic that will be published in October. The previous four1230speakers have talked about the promises of this technology. I1231want to talk about the national security implications of the1232way our country is consolidating data and feeding it to AI1233models.1234    Chairman Comer said in his statement about this hearing1235that he wanted to unleash AI responsibly while protecting1236interests and rights of all Americans. DOGE's actions do1237exactly the opposite. Over the last few months, DOGE's1238affiliates have spread across government. They are still1239exfiltrating massive U.S. data bases, processing them with AI,1240and offering them to private companies such as Palantir. These1241actions are causing irreparable harm to the security of our1242country and the safety of everyone, including everyone in this1243room, regardless of political affiliation.1244    This unprecedented security risk is a result of two things:1245DOGE's sloppy cybersecurity regarding our data and regarding1246the networks they have accessed, and also the over trusting of1247AI, current technology, to take over human tasks. You all need1248to assume that our adversaries have copies of all the data DOGE1249has exfiltrated and have established access into all the1250networks that DOGE has removed security controls from, and your1251data can be used against you. Consider the coercive power of1252your financial records or your medical records or the1253government clearance form you filled out. Whoever you think of1254as your enemy or an enemy of the United States, they can use1255this data to coerce you or a family member or a member of your1256staff, and this is true for any elected official or CEO or1257police officer or judge or nuclear power plant operator. Like,1258really, it is true for everyone. The more powerful you are, or1259the more critical your role, the more likely it is that someone1260will seek to use your data against you.1261    There is also a military concept called preparing the1262battlefield, which are the things that you do to the enemy in1263peacetime to make wartime easier. So, China hacking backdoors1264into our power grid is an example of that, and preparing to use1265our data against us is another. So, you should assume that any1266significant military action against United States will start1267with every general's bank account being zeroed out. And maybe1268your bank account is being zeroed out because there is nothing1269more distracting than your distraught families.1270    Data is power. Any entity, government or corporate, that1271holds individual data has the ability to understand, predict,1272even manipulate behavior. This is Facebook's business model.1273Our government collects incredibly intimate data about1274Americans, but its power is limited by how that data is1275organized. There was security in our data being spread among1276different government agencies with different rules governing1277use and sharing. Using AI to reason across disconnected data1278stores represents a massive increase in government power and,1279therefore, a security threat. DOGE's sloppy security practices1280also means we cannot trust the data they are using. Did they1281make copying mistakes? Did some adversaries slip in bad data.1282Remember that nonsense about 150-year-olds collecting Social1283Security? That was someone misunderstanding how the data was1284encoded.1285    Our adversaries are certainly capable of penetrating the1286security of any of the companies that DOGE gave our data to and1287just as capable at poisoning our AI systems. Data integrity is1288vital, and good AI fed with bad data makes bad decisions and is1289untrustworthy. Irreparable damage has already been done, but1290the damage is ongoing. 2015, China hacked OPM, stole data about1291all Federal workers. That was a major security breach.1292Continuous real-time access to data is much more dangerous. The1293longer these vulnerabilities persist, the more adversaries will1294be able to manipulate data and install backdoors, the more they1295will be able to manipulate the AIs we are creating, and the1296more they will be able to secretly influence our policy1297outcomes.1298    Sacrificing cybersecurity in an effort to create an AI1299future not only risks the country, it risks the AIs, which in1300turn risks our justice system, our legislative system, our1301banking system, national defense, and all of us, all of you1302personally. So, thank you, and I welcome questions and1303conversation.1304    Ms. Mace. Thank you. I will now recognize Mr. Gosar for 51305minutes.1306    Mr. Gosar. Thanks, Chairwoman. Ms. Miller, you described1307the COVID process as a catastrophic event with a high risk of1308fraud. I agree. I am the only non-U.S. resident in the United1309States to have ever killed a national emergency. I killed1310COVID, but there is not a single receipt, though, and since1311Bill Clinton through Joe Biden, we have spent over $131312trillion--let me say that again--$13 trillion without a1313receipt. Could we use AI to examine the funds spent under1314COVID?1315    Ms. Miller. It depends on where that data is available.1316Certainly, AI can be used to mine any data sets that exist. The1317question I think you are really asking is whether or not that1318data exists, and I do not have the answer to that question.1319    Mr. Gosar. Well, I hope that we would have some data base1320because we have a record of where it went, supposedly.1321    Ms. Miller. Well, yes. So, we have a record of where a lot1322of the spending went, and of course the Pandemic Response1323Accountability Committee, the organization that I helped lead,1324has a lot of that data on their website about where money was1325spent during the pandemic. A lot of that money was stolen,1326something I talk about a lot. We do not know where that money1327went. A lot of the money that was stolen during the pandemic1328went to overseas foreign adversarial nation-states.1329    Mr. Gosar. So, they also could have--how should I say this?1330My understanding, there was over 2,000 cases in Arizona alone,1331that entities or individuals were actually sued by the1332government to get it back, resulting in compensation of over1333$300 million. We could use that as well, if it disappeared?1334Once again, this is based on the data, right?1335    Ms. Miller. Yes, it is based on the data. I do not think I1336am qualified to answer that. Thanks.1337    Mr. Gosar. OK. Mr. Shah, you talked about our procurement1338system. Boy, it is broken. So, how would you use AI during that1339whole process, because our overruns in our military are1340outrageous. How can we keep them on track? How does that work?1341    Mr. Shah. Yes. I think there are a variety of things to1342talk about on that topic. One, you can use AI to help with the1343procurement process itself in terms of analyzing vendors,1344summarizing submissions, generating use cases and requirements1345for the different agencies that are looking for new technology.1346But I think before that, we need to sort of create an1347environment where these new technologies that can be applied to1348procurement or anything else in the U.S. Government can make1349their way into these agencies. And I think that is a human1350process problem that we have an opportunity here in this1351Committee to really kind of provide some oversight and1352correction so that we can go about delivering these great new1353technologies into these agencies quickly, right?1354    We have spent years working on this ourselves to try and1355get into agencies, but if you look at a lot of these1356procurement, sort of processes, they are multiyear. They are1357outdated from the 1990s where, essentially, you are thinking1358about installing software on some server rack that is going to1359last 5 years, and it might take 2 years to actually go through1360the procurement process. This is untenable for young companies1361who have limited cash, limited capital, who want to be able to1362monetize their investments and especially help the U.S.1363Government move quickly.1364    Mr. Gosar. So, could you use something like a blockchain?1365See, I am not a techie. I got questions. Could you use a1366blockchain to kind of utilize some type of--I see you. I will1367come back to you in a second, but I want Mr. Shah. Is there1368some kind of technology that will help preserve some of this1369data or save us?1370    Mr. Shah. So, we do not deploy blockchain technology for1371our solution. We focus squarely on what employees can benefit1372from when it comes to using AI to be more productive, so I am1373unfortunately not able to answer how blockchain could be used1374here. But, generally speaking, technology has been shown in the1375private sector to have tremendous impacts on their efficiencies1376and their ability to stay ahead and to lead in the industry.1377    Mr. Gosar. Yes, Ms. Miller, I got one more question for1378you. Could you use AI to audit Medicaid, Medicare? Any1379comparisons with hospital versus hospital?1380    Ms. Miller. Yes, that is a very, very good use case for AI1381because there are a lot of claims data, and there are a lot of1382information on providers, and you can look at all that data1383using AI very quickly and identify patterns that look1384suspicious. It is an area that I think really is ripe for a lot1385of innovation in the space because CMS has the opportunity to1386really mine a lot of that Medicare and Medicaid data using1387artificial intelligence to identify taxpayer dollars that are1388being stolen.1389    Mr. Gosar. Gotcha. Well, I am out of time. I yield back.1390    Ms. Mace. Thank you. I will now recognize Mr. Lynch for 51391minutes.1392    Mr. Lynch. Thank you, Madam Chair. Mr. Bajraktari, you1393mentioned in your testimony the value of educating AI literate1394workers, right? And Mr. Shah, I think you alluded to the same1395thing. About 25 years ago, I co-founded a charter school in1396Boston, a public charter school, and what we did was, we saw1397that most of the jobs that were being created were in math and1398science and STEM-related industries. So, we created this school1399that doubled or tripled the amount of educational time that1400students would be exposed to math and science. It has worked1401very, very well. Those students score very, very high on math1402and science and in the MCAS.1403    Unfortunately, on Monday, the Trump Administration1404rescinded $325 million from the STEM education grants that we1405provide to feed that type of activity, that sustain a move1406toward stronger STEM disciplines within our student bodies, and1407that was money that was already authorized, so it was not up1408for debate. This is the stuff we did last year. Cutting $3251409million in STEM grants, is that going to help our effort here1410on AI?1411    Mr. Bajraktari. Ranking Member Lynch, I think we have to be1412strategic about how we approach AI adoption at all levels of1413our educational system. I think if you look at what China is1414doing, they are introducing AI at every level of their1415educational system. I think for our kids going forward, having1416a capability in their pocket to have access to an Einstein AI1417agentic application that can help them toward tailored1418education would be an enormous benefit for the next generation1419of Americans.1420    Mr. Lynch. All right. Reclaiming my time. Learning math and1421science would probably help as well. I am going to just reclaim1422my time. Thank you though. Thank you. I think you are on the1423right track there.1424    Mr. Schneier, can you talk about--so, China has adopted an1425AI system that is total surveillance of their entire population1426there. It is non-democratic. It is quite oppressive. Their goal1427is to be able to surveil every single member of their society,1428know what they are doing, know what they are purchasing, and1429for oppressive reasons so that no one gets out of line. Can you1430talk about the need to, when we adopt, when we approve AI and1431choose the architecture for our AI, how important it is to make1432sure that transparency, that respecting privacy, independence,1433how important it is to have those elements of democracy itself1434ingrained and embodied in our AI models?1435    Mr. Schneier. It is vital, and this is why you will see a1436lot of movements toward public AI, the notion of an AI model1437that is not built and owned by a for-profit corporation, right?1438You can imagine other entities creating an AI, and that is1439really a matter of trust and integrity. Can we trust Grok or1440even the model from Google or Amazon or any of the other1441models, or do we as a country need a model that is built on1442American democratic values and not necessarily corporate1443values? It is critical to think about this. Do not think of it1444as one AI. There will be many of them.1445    Mr. Lynch. Right.1446    Mr. Schneier. I mean, already there are over a million1447models on Hugging Face that are free, and it is becoming1448cheaper to create these models.1449    Mr. Lynch. In your testimony, you talked about the dangers1450that Elon Musk created by the way he was diving into Social1451Security data and IRS data, and the danger that that presented1452to the United States from a national security standpoint and to1453the privacy of U.S. Citizens. Can you talk about that, please?1454    Mr. Schneier. I mean, it is really a matter of process. The1455goal might be a good idea, but the reason we have controls1456around this data, who can access under what rules in what1457configurations, are because it is so sensitive. You think about1458the different data enclaves, whether it is census or military1459data or OPM data about government employees or health data.1460This is very critical, and we, as citizens, give this data to1461the government for good reasons. And it is kept separate for1462good reasons, and it is kept secure. So, when we read about1463laptops of data being removed, data being given to Palantir to1464train models that they are going to use to sell back to us, the1465security is a real concern.1466    Mr. Lynch. Madam Chair, I yield back. Thank you.1467    Ms. Mace. Thank you. I will now recognize Mr. Higgins for 51468minutes.1469    Mr. Higgins. Thank you, Madam Chair. Mr. Thierer, I am1470going to be talking with you mostly for the next 5 minutes1471because I would like to dig into the controversy regarding the1472moratorium section of the BBB legislation currently being1473considered by Congress. It is across the country, as part of1474the narrative of the discussion of AI, and I believe we have an1475opportunity today to clarify the intent and the specifics of1476that section of legislative language in BBB. So, in that topic,1477and please help me to understand, and, in doing so, help the1478Nation to understand, good sir.1479    My read of the actual bill states, to that section in1480particular going to the moratorium under the section of1481artificial intelligence and information technology1482modernization, it says in general, except as provided in1483paragraph 2, which is essentially an exception to the1484moratorium that it allows for sovereign states to enact and1485enforce laws that remove restrictions on use of AI. What the1486paragraph says in general, except as provided in Paragraph 2,1487no state or political subdivision thereof ``may enforce''1488during the 10-year period, beginning on the date of enactment1489of this bill, any law or regulation of that state.1490    So, my read is that the state is not restricted from1491debating and considering, and voting on and passing, and using1492their legislative bicameral body in each sovereign state to1493debate and vote on and pass laws regarding AI. And for that1494Governor to sign that law, the Big Beautiful Bill in this1495section states that the Federal Government is asking for a1496pause on enforcement. Am I right or wrong, and please1497embellish.1498    Mr. Thierer. Yes. Thank you, Congressman for the question.1499So, I testified about this in front of the Energy and Commerce1500Committee just 2 weeks ago where there was a debate about the1501AI moratorium. And it is important to keep in mind that the1502reconciliation bill includes specific funding to modernize1503Department of Commerce technology systems, and there is a1504concern among Members about exactly how that can happen if1505there is a host of conflicting laws at the state level.1506    Mr. Higgins. Right, and the so-called moratorium is under1507that section.1508    Mr. Thierer. That is correct.1509    Mr. Higgins. Please continue.1510    Mr. Thierer. And so, therefore, that would prohibit certain1511AI-specific types of regulations at the state and local level1512again----1513    Mr. Higgins. It would prohibit the enforcement of state-1514authored regulations.1515    Mr. Thierer. That are AI-specific regulations. It very1516specifically says that laws of general applicability are not to1517be covered by this, also criminal activity not covered. And so,1518there are some important exceptions to this, and states are1519going to be free to continue to act on a lot of these different1520things.1521    Mr. Higgins. In the interest of time, just to clarify for1522the country, sir, this moratorium does not restrict a sovereign1523state's right to debate, vote on, and enact new AI legislation1524within that state. The moratorium is asking the states to1525recognize that AI, by nature, is interstate commerce, and under1526this section for modernization of the Commerce Department of1527the reconciliation bill, we are stating that the states should1528refrain from enforcement of new AI laws. Is that correct?1529    Mr. Thierer. Yes, Congressman, and specifically, what you1530said about protection of the interstate, the national1531marketplace, is really crucial, and it reflects the broader1532consensus of Congress going back over the last 30 years about1533the internet and technology more broadly, that there does need1534to be somewhat of a national framework and make sure that we do1535not have a parochial patchwork that will interfere with the1536free flow of algorithmic commerce and speech.1537    Ms. Mace. Six seconds.1538    Mr. Higgins. OK. That was a squared-away explanation. I1539thank the gentleman. Madam Chair, I yield.1540    Ms. Mace. Thank you, Mr. Higgins. I will now recognize Ms.1541Norton for 5 minutes.1542    Ms. Norton. Thank you, Madam Chair. Government must1543leverage the best technology possible to meet the ever-changing1544needs of the people it serves. From the creation of the1545national Postal Service, to the discoveries that led to the1546internet, the Federal Government has a long history of1547embracing new technology to better deliver for our1548constituents. But our North Star must always be service to the1549American people, not tax breaks for billionaires or back-room1550deals to benefit the richest man in the world at public1551expense. As we discuss the use of artificial intelligence in1552government, we must have people at the center of the1553conversation, including the people we represent and Federal1554workers who serve our country. AI is a tool that can empower1555our public servants, but it is not a substitute for passionate1556and committed Federal workers.1557    Ms. Miller, can AI alone replace Federal workers?1558    Ms. Miller. Absolutely not, no. The implementation of AI,1559it is very important when we think about building AI tools,1560that we are building them with the people that we will use1561them. AI is talked about a lot as something that is going to1562replace people. I am a strong advocate in using AI to automate1563routine processes and repeatable processes and free up people1564to do more high-level, high-skilled work. And so, I absolutely1565do not believe that AI should be doing the work of the public1566servants. I believe that AI should augment that work and that1567we should work with public servants as we put in place AI1568tools.1569    Ms. Norton. Thank you. Elon Musk and the Department of1570Government Efficiency has gutted so much of the Federal1571workforce, firing tens of thousands of people under the guise1572of greater efficiency, but Americans know better. They see the1573negative impacts every day. As a direct result of the1574interference with Social Security Administration, senior1575citizens now face 98-minute wait times to speak with customer1576service, crowded lobbies at field offices, and being turned1577away from scheduled appointments because the Agency simply does1578not have enough employees for its mission. Poor customer1579service is not efficiency. It is failing to uphold the1580government's responsibility to our constituents. Now the Social1581Security Administration wants to cut an additional 7,0001582employees and somehow use AI to handle calls from seniors. That1583does not work because I know that when my constituents have a1584problem with Social Security, they would rather talk to a1585person than a computer program.1586    The Department of Government Efficiency cuts are wreaking1587havoc on our communities, and it will take years to clean up1588the mess Elon Musk has left behind. Thank you, and I yield1589back.1590    Ms. Mace. Thank you. I will now recognize Ms. Foxx for 51591minutes.1592    Ms. Foxx. Thank you very much, Madam Chairman, and thanks1593to our witnesses for being here. Mr. Thierer, the Trump1594Administration has expressed a goal for the United States to1595maintain its dominance in AI. Why is it important for the1596United States to maintain its dominance position in AI?1597    Mr. Thierer. Thank you for the question, Congresswoman. As1598you have heard here today from several witnesses, we are in a1599serious race with China and the Chinese Communist Party for AI1600supremacy globally. And the broad-based ramifications of who1601wins that race are significant for not just our geopolitical1602security, but also our values broadly as a Nation and for the1603world. And so, there is a real danger that if we do not get AI1604policy right as a Nation, that China will take the lead. Its1605products will diffuse throughout the world faster, and we are1606already seeing this. I was testifying earlier this year at a1607hearing on the so-called DeepSeek moment, and it surprised a1608lot of us, including me, how advanced these Chinese systems are1609today and how effective they have been at filling gaps globally1610that America is not there to fill at the moment.1611    Ms. Foxx. Thanks. I do think the answer is pretty obvious,1612but it was important to get something on the record.1613    Mr. Thierer. Sure.1614    Ms. Foxx. How has the Trump's Administration approach to AI1615differed from the previous Biden Administration's, and why is1616the departure from the Biden Administration's AI policy1617necessary, Mr. Thierer?1618    Mr. Thierer. Well, the Biden Administration approach was1619very voluminous in terms of rulemaking, obviously, executive1620order, longest in American history on AI, over 110 pages. And a1621lot of it was very fear based, unfortunately, and did not1622embrace the technology as much as the new Trump Administration1623executive order and subsequent orders have done. And so, I1624think there has been a real sea change of approach here in1625terms of understanding the benefits associated with AI1626technologies broadly.1627    Ms. Foxx. Thank you. Mr. Shah, your company has been1628successful at streamlining certain administrative tasks for1629businesses in the private sector and helping boost1630productivity. How do you see AI technologies like those1631developed by your company being effectively used in the Federal1632Government, and what kinds of productivity gains and cost1633savings could we expect to see?1634    Mr. Shah. Thank you for the question. You know, to put it1635in maybe legislative terms, imagine you had an assistant that1636could summarize the status of a bill. Perhaps maybe look up a1637constituent problem that you have been tackling before a1638meeting or be able to search all of your notes and your memos1639about various Committee hearings and tasks that have been1640assigned. All these things are things that AI gives us today,1641and the private sector is using this quite aggressively to make1642their own teams more efficient and effective.1643    To your question about sort of what that looks like, you1644know, there are sort of three categories I like to think about.1645Scale. Broadcom, for example, was able to go from 10,0001646employees to 50,000 while keeping the same support staff. So,1647that is another benefit, which is you can take the same team1648and allow them to support a lot more employees in this case,1649but in our case here today, talking about constituents. I think1650there are also employee time savings. The city of Glendale I1651mentioned did a 500-percent plus ROI on their investment, but1652that was 3,500 hours of savings from the employees themselves1653that they could apply and use toward other tasks and functions1654for their constituents. And then, of course, there are monetary1655savings, which is that if we can do some of this more1656efficiently, that goes back to the U.S. taxpayers. It goes back1657into the economy to get spent on further innovation.1658    Ms. Foxx. Thank you very much. Ms. Miller, according to the1659Office of Management and Budget, each year, taxpayers provide1660more than $1.2 trillion, or nearly 5 percent of GDP, in funding1661for thousands of programs across the entire government through1662grants and other financial assistance. With so much taxpayer1663money spent, it should come as no surprise that, besides a very1664high-level accounting, the Federal Government largely cannot1665tell whether a particular grant award was used wisely or if it1666accomplished the stated goals of the grant program. This1667Committee has worked on legislation such as my bill from 2019,1668the GREAT Act, to address the problem, but as we have1669discovered, there is not enough data available in the proper1670formats to evaluate whether taxpayer funds given out as grants1671are being used wisely.1672    How could AI help us solve this problem and make sure that1673every taxpayer dollar given as a grant is used effectively and1674for its stated purpose?1675    Ms. Miller. Thanks for the question. Actually, we did1676something pretty innovative with my technology company,1677TrackLight. We were trying to figure out who one of the1678agencies that we were working with paid, and so we got access1679to minutes of board meetings, and then we used AI to scrape1680those minutes and identify vendors. And then we were able to1681collect hundreds of vendors that this agency was paying that1682they did not know they were paying. Now, we can all talk about1683how insane it is that the agency does not know who they are1684paying, which is a serious problem. But there are ways that AI1685can help solve this problem and start to, sort of, piece1686together who these grants are going to, and it is incredibly1687important for taxpayers to know that information.1688    Ms. Foxx. It would be nice if we could just search things1689on fireside. Thank you.1690    Ms. Miller. Yes.1691    Ms. Mace. Yes. All right. Thank you, and I will now1692recognize Mr. Mfume for 5 minutes.1693    Mr. Mfume. Madam Chairwoman, thank you very much. My thanks1694to the guests who have come today. One of the things that1695strikes me about this hearing is that we do not have enough1696time, and we are all trying to get a lot in here. So, I will1697certainly use the right that Members have to submit questions1698for the record for 5 business days because there are a number1699of you. Mr. Shah, in particular, I want to ask some questions1700of you through that process, and, Mr. Schneier, your testimony1701was chilling to say the very least. So, hopefully, we can talk1702about that, but, Ms. Miller, I would like to start with you.1703    Much of your work focuses on fraud prevention, and before1704we get to that, I would like to go back to something that I1705heard that was just as chilling, and that was your description1706of these outdated privacy laws, which do not afford any1707protection whatsoever to what is happening now in 2025 at the1708rapid pace of this technology and its development. It is very1709scary to believe and to hear, quite frankly, that we are behind1710in that because the privacy of Americans is absolutely1711important. It is just as important as what Ms. Norton brought1712up earlier, and that is the fact that people who are looking1713for Social Security or trying to get Medicaid or trying to1714figure out SNAP benefits are running into this technology and1715being run over by it in many instances. So, can we talk just a1716minute about these privacy laws, which I assume they predate1717all of this discussion, and that there is, as I understand, not1718a real determined effort to upgrade them to be able to protect1719industry, and to be able to protect government and individuals?1720    Ms. Miller. Yes. Thank you, Congressman. The Privacy Act1721was passed in 1972, which was 52 years ago, almost 53. That is1722still the law that governs how we protect privacy today, which1723obviously is exceptionally outdated. In my opinion, there is1724not a lot that can be done through the legal system to protect1725Americans' personally identifiable information anymore. I think1726we just live in an entirely new world now, and foreign1727adversarial nation-state actors, criminal rings, all of our1728data has been monetized, is being monetized today. Ninety-six1729percent of information that is stolen through a data breach is1730used to monetize by a threat actor.1731    And so, because we have such an asymmetric situation when1732it comes to data, government is going to have to get much1733better about using technology to protect privacy. And there are1734technologies that are in place, I am sure, that some of the1735panelists here today are using in the private sector. We can1736anonymize data. And so, we can do things that can protect1737people's privacies, and we are going to need to think about1738this both from a legislative perspective, not hamstringing1739agencies. The Privacy Act and the Computer Matching Act keep1740agencies from being able to share data that they desperately1741need to be able to find these actors, and, again, these actors1742are foreign adversarial actors.1743    Mr. Mfume. Thank you very, very much. Mr. Schneier, let us1744go back to your assessment of the bigger you are, the bigger1745target you are, the more data is collected on you, and the more1746harm that will come to you, your community, your agency, your1747government in this instance. Walk me briefly down that street1748again because you did not go into the fact that this is not1749science fiction. This is happening every day, and it is1750happening to the United States of America.1751    Mr. Schneier. I mentioned China, the OPM breach from the1752last decade where China went into OPM and took the data, among1753other things, all of the clearance forms of United States1754citizens, and you can imagine why they might do that, and why1755they might identify who spies are, might look for people they1756can influence, and that is the sort of thing that countries are1757doing today. I mean, I am assuming we are doing it to them, and1758so this data is very valuable at the government level for a lot1759of different things.1760    I mentioned two scenarios. I mentioned a potential war1761scenario, potential peace scenario, and a lot of--our data is1762being collected everywhere. And so, I mean, a lot of data is1763had by social media companies that know a lot about what we1764like, and who we are and who our friends are, and who is1765important to us, but the data we give governments is a1766different sort of data If you think about the data that is on1767the tax return or the data that the Census Bureau collects, or1768the military data, that is personal on a different level, and1769countries want that. They are going to want that in aggregate1770to figure out things about us, and they are going to want it1771individually. And coercion means--lot of it depends on who you1772are, what coercion means.1773    Mr. Mfume. Thank you. My time has expired. Madam Chair,1774thank you very much.1775    Ms. Mace. Thank you. I will now recognize Mr. Sessions for17765 minutes.1777    Mr. Sessions. Madam Chair, thank you very much, and good1778morning to each of the panel members. We appreciate you being1779here.1780    Long time ago, I spent time at Bell Labs where we were1781responsible for the network architecture but also the1782development of the switch that would be utilized. We set1783standards and did things to where we had unified way to look at1784how switch operated. We both know we had Salt and Volt Typhoon1785whereby our government people learned that the Chinese had been1786into, through cyber, our data bases for a long time. Should1787have found it. Did not find it. It cost us. My question really1788is, and I am sure every one of you have an answer for this or1789at least your vision of that. All it takes is one node to be1790compromised. It is like a big fence that goes around. It just1791takes one loose place.1792    Does the Federal Government, do we here in Congress need to1793ask someone? Do we need to deal with other countries, that1794there need to be a standards body that specifically, like when1795I worked at the Labs, to set what those standards are? How do1796we avoid, with all the things we are doing, getting something1797that happens like that again? Sir, evidently you have got your1798hand on the panel first. Any ideas of that?1799    Mr. Schneier. I did not know this was Jeopardy, that you1800got to be quick.1801    Mr. Sessions. Well, OK. Then we would go to you last if you1802want.1803    Mr. Schneier. No----1804    Mr. Sessions. But my point is, do you have any ideas on1805these question?1806    Mr. Schneier. So, we have things. I mean, this is the job1807of the NSA, right? We have security organizations in the United1808States whose job is to protect American data, and I think we do1809actually do a pretty good job. And we have problems that a lot1810of data is in private hands, that, you know, right now, if you1811think about just the Defense Department, there are thousands of1812networks, but there are rules, and, you know, I think we are1813doing OK. I do not think we need an international standard. I1814mean, this is very much United States. I mean, NATO does things1815together, so there are groups, but this is not like a corporate1816security thing where you need an international standard.1817Everyone is consistent. The U.S. has U.S. problems, U.S.1818adversaries, and we do have U.S. solutions.1819    Mr. Sessions. We also have battleships and aircraft1820carriers that are out in the middle of places. Anyone else?1821Thank you, sir. Mr. Shah?1822    Mr. Shah. Yes. I think, you know, what we have seen,1823especially in cybersecurity, the NIST has created standards1824there that became adopted by a majority of the industry. And I1825think when it comes to AI and protecting data, coming from the1826lens of a startup, from a young company, we have looked at the1827new standards that they have around AI and they are actually1828pretty good, and they do provide for a lot of recommendations1829that we actually follow. And then furthermore, you know, going1830through the FedRAMP process, while I found it to be very1831expensive, there was a lot of amazing sort of things that they1832recommended in terms of infrastructure, software, process, and1833procedures that allow us to really make sure that the data is1834secured. And so, I think that if we can have more of these1835types of frameworks, more of a lighter touch, it will allow us1836to continue to innovate fast but still give us all a true north1837of where we should take things in terms of our infrastructure1838and our processes.1839    Mr. Sessions. Sir?1840    Mr. Bajraktari. Sir, I would say three things. No. 1, just1841to follow on Mr. Shah's point, I think NIST is well positioned1842for this. I think the re-missioning of the Safety Institute by1843the Department of Commerce 2 days ago, that will focus on1844really building the standards across our Federal Government and1845how they use AI, is a step forward. Also, looking at, you know,1846open-source models coming from countries of concern,1847specifically China, and what kind of a risk they possess in our1848ecosystem, I think that is the right step to do it. At the1849allies and partners levels, as I mentioned in my opening1850remarks, The Five Eye Alliance really provides us with a1851stepping stone of, you know, coming with some common standards1852among our five ally partners and then start bringing other1853countries under the fold. And the third piece, as Mr. Schneier1854mentioned here earlier, what can we do on the offensive side so1855our adversary does not get a hold of our data, they do not use1856it against us, and then actually we put them on the back foot1857instead of letting them, you know, really attack us on a daily1858and hourly basis.1859    Mr. Sessions. Good. Anyone else?1860    Mr. Thierer. Just very briefly. Sir, I just want to put in1861a good word for what the Trump Administration has been doing1862with its recent executive orders and OMB guidance on this1863front, which has been to provide not only a framework for1864expanding the use of AI throughout government, but also getting1865more serious about security vulnerabilities and other types of1866concerns, and recent OMB guidances like M-25-21 and 22,1867basically try to address these concerns you are raising.1868    Mr. Sessions. Great. Thank you. I appreciate the panel very1869much. Madam Chair, I yield back.1870    Ms. Mace. Thank you. I will now recognize Ms. Brown for 51871minutes.1872    Ms. Brown. Thank you, Madam Chair. I am glad this Committee1873is finally turning its attention to the Federal Government's1874use of artificial intelligence because under President Trump,1875AI is not being used to serve the people. It is being abused to1876serve the powerful. DOGE was a reckless, unregulated, and1877unethical failure, and I would not be surprised if some of my1878colleagues on the other side quietly agree now that Elon Musk1879is out here trashing their one big ugly bill.1880    So, let us be clear about how we got here. Musk and his1881cronies were handed the keys to our government's data, let1882loose to sidestep security safeguards, strip away Federal jobs,1883and exploit Americans' personal information unchecked and1884unaccountable. They used AI as a blunt force tool to slash1885thousands of Federal jobs and gut billions in government1886spending with no transparency or regard for the human cost, but1887that is not even the most dangerous part. DOGE has launched an1888unprecedented power grab over Americans' personal data. Earlier1889this year, reports surfaced that a DOGE staffer, with no1890credentials and no clearance, reportedly accessed Americans'1891personal information and shared private data with third1892parties. I co-led a letter with Reps Trahan and DelBene1893demanding an investigation, and it might not come as a surprise1894that we are still waiting for answers.1895    And we now know this was not a one-off. Whistleblowers have1896revealed that DOGE is actively building a master data base, a1897centralized system combining sensitive data from across Federal1898and state agencies that includes Americans' medical records,1899their credit card numbers, their medical claims, their student1900debts, and much, much more. To build it, the Administration has1901turned to Palantir, the surveillance firm funded and founded by1902Republican MAGA donor, Peter Thiel, and this massive trove of1903personal information is being built with no oversight, no1904security guarantees, and no guardrails.1905    Here is why every Americans should be alarmed. In the wrong1906hands, this data base could be weaponized to target political1907enemies, suppress dissent, or surveil the public, and even in1908the best-case scenario, a breach could expose millions to1909identity theft, blackmail or worse. What is most disturbing is1910how much we still do not know, but what we do know raises1911serious concerns about violations of the Privacy Act, FISMA,1912and E-Government Act. This Committee must keep investigating1913because we cannot allow AI to become a weapon for surveillance,1914exploitation, and the dismantling of public institutions.1915    So, Mr. Schneier, how should Americans feel knowing that1916DOGE is pulling together their most sensitive personal data to1917train AI without clear safeguards, oversight, or consent, and1918what protections are in place to keep that data from being1919misused or exposed?1920    Mr. Schneier. Well, we do have a lot of laws in place. We1921have laws protecting the data. We have laws prescribing how it1922is used, how it is shared, and that is what is being bypassed.1923So, it really is more a matter of enforcing the laws we have on1924the books than writing new laws. I do not think we have a need1925to do anything because of AI because these problems with data1926sharing/security risks have been with us for decades. And we do1927have laws, we do have rules, and they are being bypassed.1928    Ms. Brown. So, let me ask you this. Given DOGE's reported1929practice of bypassing security protocols and exporting1930sensitive data to less secure environments, how serious is the1931national security risk, and are we making it easier for foreign1932adversary hackers to access government systems and Americans'1933personal information?1934    Mr. Schneier. I mean, we are. We knew from the first weeks1935that some things DOGE was doing was bypassing security1936protocols and removing audit trails. I mean, basically1937eliminating the security mechanisms we have in place to protect1938the data. We had one example of a Russian actor going in after1939a DOGE employee using their credentials, and that is the one we1940know about. These countries are not stupid. They know how to do1941espionage. They have been doing it for decades. And so, I think1942what I said in my remarks, we have to assume that all of this1943data has been exfiltrated by our enemies. I would be very1944concerned. Now, it depends on who you are, the more powerful1945you are, the more concerned you are because more leverage can1946be leveraged against you.1947    Ms. Brown. Thank you. Let me be clear. Artificial1948intelligence and other emerging technologies can offer real1949benefits but only when developed responsibly and with clear1950guardrails that protect the American people. So, I stand ready1951to work with my colleagues on the other side of the aisle who1952are serious about securing our Nation's cybersecurity, safely1953deploying AI within the Federal Government, and ensuring that1954Americans' data remains safe. And with that, Madam Chairwoman,1955I yield back.1956    Ms. Mace. Thank you. I will now recognize Mr. Biggs for 51957minutes.1958    Mr. Biggs. Thank you, Madam Chair. Thank the panelists for1959being here today.1960    Mr. Shah, in your written testimony, you describe how1961various communities, counties, cities, and two of them in1962Arizona--Scottsdale and Glendale--are actually using AI,1963perhaps becoming a hub of AI, anyway, using to become more1964efficient. Can you expand on that for us? Tell us what they are1965doing and the guidelines or guardrails that they have in place.1966    Mr. Shah. Yes, it is a great question. So, you know, in1967terms of the cities and counties that have started to adopt1968this technology, they are using it to make their employees more1969effective. What does that mean? That means that they can self-1970serve a lot more use cases than they would have otherwise. They1971can find information. They can troubleshoot issues that they1972might be having with IT or even an H.R. situation or use case.1973They can submit performance reviews. They can do a variety of1974tasks that were otherwise very manual and very slow, and1975required a lot of coordination amongst folks.1976    And so, you know, as I was mentioning earlier, the city of1977Glendale in Arizona is saving their employees 3,500 hours now1978each month, being able to now do more with those same team1979members to support the employees or the constituents of their1980county. And so, I think that we see this across the private1981sector. We have seen people use this technology to enable more1982employees to do greater levels of work and greater levels of1983output, while still sort of maintaining all the privacy and the1984security that is required in any of those environments.1985    Mr. Biggs. Thanks for that. Now, Mr. Schneier was talking1986about the laws that have been placed to protect privacy rights,1987and they have been in place for a long time, largely ignored1988from both parties, quite frankly, in my opinion. But what I1989want to get to is something that Mr. Bajraktari said, when you1990start talking about The Five Eyes community. And I just came1991from another hearing where one of the issues that we are having1992is that, pursuant to the CLOUD Act, the U.K. is trying to get1993backdoor hacks into Apple, and they are claiming that they have1994some kind of robust privacy protection there. But the reality1995is they will not disclose what that robust privacy protection1996is, and they are keeping the entire process closed, obscured.1997And so, I am curious about how we can fully work with The Five1998Eyes community in the development of AI and yet protect the1999fundamental rights of American citizens, and so I am going to2000ask you that question.2001    Mr. Bajraktari. Thank you, sir. Great question. We created2002The Five Eye community during the cold war because one of the2003ways we could share more and fast and be ahead of what Soviet2004Union was doing was, you know, being able to share with our2005allies and partners. I think there needs to be a starting point2006in how we confront China today. I do not think we can confront2007China alone without bringing, you know, our closest allies2008under the fold. The speed and scale of how China adopts and2009uses AI for national security purposes is really breathtaking.2010So, my point is, we should start with a community of the most2011trusted allies and partners that we have.2012    Mr. Biggs. So, my question is, though, when you have,2013essentially, the abuse of rights that we are seeing through the2014abuse of the CLOUD agreement between U.S. and U.K., can you2015fully trust The Five Eyes to work together on this? And that2016is----2017    Mr. Bajraktari. Yes, I understand. I understand your2018concern, sir. I think we have to work really closely with them2019to remove obstacles because I think the challenge we face ahead2020of us is that, if we do not find a common language and build2021common security protocols, I think we would have a different2022world by the end of this decade, you know, a world in which CCP2023will probably have access to all our data. The entire global2024infrastructure will depend on the CCP and the companies2025subsidized by Chinese, so we have to find ways to remove these2026obstacles with our closest allies.2027    Mr. Biggs. I do not disagree with that. It is just, we need2028to make changes right now, and we are not making changes fast2029enough, and we are not implementing anything fast enough.2030    So, I am going to ask unanimous consent to enter into the2031record the bill text from a new Arizona law passed earlier this2032month to close dangerous loophole that allowed these predators2033to use AI and digital editing tools to create child sexual2034abuse material that appears real, but depicts no actual child.2035It was HB-2678. It is sponsored by House Republican Majority2036Whip, Julie Willoughby, and it updates state law to ensure AI-2037generated and digitally altered images that are2038indistinguishable from real minors carry the same criminal2039penalties as those involving actual children. I would like to2040submit for the record.2041    Ms. Mace. No objection.2042    Mr. Biggs. I also ask for unanimous consent to enter into2043the record an article discussing how my home state of Arizona2044is taking steps to allow AI to flourish, showcasing that states2045can handle AI regulation responsibly and reap the economic2046benefits.2047    Ms. Mace. No objection.2048    Mr. Biggs. Thank you.2049    Ms. Mace. All right. I will now recognize Ms. Stansbury for20505 minutes.2051    Ms. Stansbury. All right. Good morning, everyone. Thank you2052so much for being here, and I want to say thank you to the2053entire panel for your recommendations this morning. My2054background is actually in the sciences. I have actually worked2055on natural and earth sciences most of my career, including on2056Big Data and modeling, so I do appreciate much of what the2057recommendations are that have come from our experts.2058    But I am especially grateful that you are here, Mr.2059Thierer, because when the huge abomination of a bill, I will2060call it, came down the pike, my eye actually caught this2061specific language and funding appropriation in the bill that2062you referenced in your testimony. And I think many of us were2063confused about where it came from, what its intent is, and this2064is the first time I have had the opportunity to actually ask2065some questions, so I am glad you are here. And as I understand2066it from my look into your background, that you have actually2067been directly involved in developing this language and request.2068Is that correct?2069    Mr. Thierer. Not that language. I have written about2070moratoria in the past, though, going back to the 1990s when I2071worked with Congress on the Internet Tax Freedom Act.2072    Ms. Stansbury. Got it. OK. So, you mentioned that part of2073why this deregulation language is in here that would preempt2074state and local laws regarding the regulation of AI systems is,2075in part, because of this $500 million request that is in the2076bill. Can you help us understand what would that $500 million2077be used to do specifically?2078    Mr. Thierer. Going back to the Bipartisan House AI Task2079Force report from last December, there was a major section that2080led off the report about the need for government modernization2081of IT.2082    Ms. Stansbury. Just some specifics.2083    Mr. Thierer. So, specifically, at the Department of2084Commerce, there is a need to actually improve how various2085programs function there, including new ones that have just been2086created.2087    Ms. Stansbury. Just some specifics, like, what kind of2088systems would be built with this $500 million?2089    Mr. Thierer. Well, I do not know exactly. I do not know the2090answer to that, Congresswoman. I am sorry, I do not know2091specifically what Commerce will use that money for going2092forward.2093    Ms. Stansbury. Do you know where this request came from?2094    Mr. Thierer. You would have to ask the House Energy and2095Commerce Committee specifically who put that language forward2096in the----2097    Ms. Stansbury. Do you know what companies would benefit2098from this language?2099    Mr. Thierer. No, but I do know that there is a broad need2100for many different types of vendors and services to be utilized2101by the Federal Government, not just----2102    Ms. Stansbury. So, this would likely go to private2103contracts, correct?2104    Mr. Thierer. It would go to many private contracts,2105certainly.2106    Ms. Stansbury. Yes. OK.2107    Mr. Thierer. Yes.2108    Ms. Stansbury. And so, and some of the leading contractors2109who do work in this area that already have government contracts2110are companies like Palantir, correct?2111    Mr. Thierer. I believe that is one of many, but there are2112many other vendors of IT services and AI.2113    Ms. Stansbury. So, correct, yes. And I heard you argue just2114a few moments ago when Mr. Higgins was asking you about this2115preemption language, that it was needed because there would be2116integration of data from state and local entities. And it is my2117understanding--I think others have kind of alluded to this--2118that right now, the Trump Administration is working on creating2119some sort of master data base that brings together data from2120multiple different sources across state lines. And in fact,2121there was a request that went out to the states asking for data2122about SNAP beneficiaries because my understanding is Homeland2123Security is interested in integrating data for immigration2124enforcement purposes. So, they would like to integrate data2125using the census data that they hacked in, exfiltrated, and2126downloaded. They would like to integrate data with the2127immigration data bases that they have, and now they are asking2128states to send data that falls under state jurisdiction to2129integrate for immigration enforcement.2130    And, you know, you can call it whatever you want. I have2131heard folks say this is not a master data base, but it2132absolutely is, and we know that the Trump Administration is2133trying to take American data and hire private companies like2134Palantir to create means to use American data for their own2135policy ends, including endangering the lives of Americans. So,2136I think that one of the reasons why folks are reacting so2137strongly to this language that is in this bill is because it is2138so extreme. And yesterday, I heard Marjorie Taylor Greene, I2139was standing on the Floor while she gave a speech on this very2140language that is in this bill that would literally say that2141states and localities cannot enforce their own data rules or2142AI. It preempts federalism. It preempts state law. It would2143prevent states from protecting their own citizens.2144    And you know, it is rare these days that we have bipartisan2145agreement, but I agree with literally everything that Marjorie2146Taylor Greene said on the Floor yesterday. This is dangerous.2147It could endanger American lives, and we cannot allow this2148language to get through. And with that, I yield back.2149    Ms. Mace. Thank you, and I will recognize Mr. Perry for 52150minutes.2151    Mr. Perry. Thank you, Madam Chair, and, ladies and2152gentlemen, thanks for making the trip and being here. I am not2153sure what questions I have for you, but I do want to make2154everybody aware of some things that I am aware of. I am going2155to refer directly to an AI safety firm named Palisade Research,2156and another one named Apollo Research and talk about things2157like the fact that the Chinese Communist Party likes to pollute2158the, I do not know, the available information in the world with2159all kinds of things that support the Communist Party of China's2160view of the world, and not the free world's.2161    And with that having been said, I suspect that all of you2162are aware of the times that when various AI has been tested. I2163will give you a couple names--Codex-mini, OpenAI model o3,2164OpenAI model o4--and other ones, the AI has chosen, on2165occasion, to further its goals, as opposed to the goals of the2166people that wrote the code. And on occasion, how about the2167instance where Anthropic's Claude 4 model attempted to2168blackmail the people it believed were trying to shut it down,2169and then it would ignore and other ones would ignore shutdown2170commands, even when told in advance that shut down commands2171were coming, and went as far as to rewrite itself and preserve2172itself on a different server to avoid the commands of the2173humans in charge of it. Now, that might seem like the view of a2174Luddite, but, I mean, I have got pages and pages of AI2175deliberately disabling various developer-installed oversight2176mechanisms, shutdown commands being ignored, replicating2177itself, preserving itself without the user even knowing.2178    Now, let me be clear. I think that the United States of2179America has to win the race and be preeminent in AI, but I also2180think, while I sit on the Oversight Committee, that knowing2181these things are occurring during tests, that us encouraging,2182as a committee or as a Congress, the use of these systems in2183places like nuclear security or national security without2184having these pretty obvious things worked out is dangerous,2185like, at a minimum. I cannot even believe we are having the2186conversation. Like, how in the hell are we in here telling2187everybody that we have got to incorporate this into the Federal2188Government, state government, all levels of government,2189including, like, National Defense, nuclear security, when we2190know these things are occurring?2191    And I do not know which one of you to ask. You all seem to2192be all for AI, and I get it is your vocation, your passion, et2193cetera, but we got a duty here on this Committee and in this2194Congress to make sure that the right thing happens. And you2195know, humans, Americans, are supposed to be in charge of the2196government, not some machine that learned from another machine2197that picked up data that was polluted with Chinese propaganda2198and is going to, like, push the button for us. Ms. Miller, you2199seem interested. I got a minute.2200    Ms. Miller. Yes.2201    Mr. Perry. What do you got?2202    Ms. Miller. Well, this is a real concern, and GAO and other2203standard-setting entities have been really pushing for2204responsible AI frameworks. And this is one of those reasons2205why, when we talk about AI, you can get a bunch of people in a2206room that are like, go, go, go, lean in, you know, do more2207adoption faster, faster. But it is so important to think about2208these safeguards and how important they are. Responsible AI is2209incredibly important when we are dealing with national security2210information and American citizens' information, things that2211could threaten the actual, you know, safety and security of2212American citizens. And so, it is incredibly important that we2213are slow and deliberate enough, and that is one of the reasons2214why GAO and others have put out these frameworks. And it is2215really important, I agree with you, as this Oversight2216Committee----2217    Mr. Perry. So, are you saying that the frameworks deal2218adequately with these things, or what am I missing?2219    Ms. Miller. Well, I am saying that if they are considered2220and carefully rolled out, then those concerns will be at least2221taken into consideration. I do think there is a lot of risk in2222AI, and we should, I agree with you, consider where we want to2223apply this and where we want to be----2224    Mr. Perry. So, ma'am, I have a little bit of time left, and2225I just want to say this. I think, for me, it is beyond2226consideration. Like, before we hand this thing the nuclear2227codes, I do not want it to be considered. I want to know the2228damn stuff is right, not polluted with Chinese Communist facts2229or their alleged version of the facts and that the thing is not2230going to avoid the humans and avoid, you know, detection and2231all that stuff.2232    Madam Chair, I would like to enter into the record or2233request that be entered into the record, Apollo Research's2234report titled, ``Frontier Models are Capable of In-Context2235Scheming,'' published on January 16, 2025, and I ask that2236actually----2237    Ms. Mace. No objection.2238    Mr. Perry. Thank you. I yield.2239    Ms. Mace. Thank you. I will recognize Ms. Randall for 52240minutes.2241    Ms. Randall. Thank you, Madam Chair, and thank you to our2242witnesses today for being here, maybe longer than you even2243planned.2244    You know, Mr. Perry mentioned some of the important reasons2245that AI use in the Federal Government is responsible for the2246lives and livelihoods of so many more individuals than, you2247know, one of us asking ChatGPT a question, you know, for our2248own purposes. And we have seen some examples of potential2249Federal Government uses of AI that have gone wrong. Mr.2250Schneier, you know, Elon Musk and DOGE have tried to throw AI2251at problems throughout the government with no regard to2252Americans' privacy, safety, and security, no regard to the225327,000 Federal employees who I represent and do not seem to be2254able to parse the difference between gender research and2255specifying that female flies must be used in NIH research2256studies. Can you discuss why the need for some of those higher2257standards in, you know, a higher standard of AI use framework2258in situations like that, and what some of the pitfalls are of2259not providing a higher standard for AI use in government than2260the private industry?2261    Mr. Schneier. I am not sure it is government versus private2262industry. I mean, we probably might expect the same standards2263as, like, banking or in university admissions, right? There are2264going to be these instances where the decisions made are2265important enough that we have rules and laws prescribing how2266they can be made, and it is really a matter of making sure the2267AI is suitable for the task, being able to audit what the AI is2268doing, correct for mistakes. You know, humans make mistakes all2269the time, and we as a society has had thousands of years2270dealing with human mistakes, from double entry bookkeeping to2271checklists of surgical equipment, make sure they are not left2272in bodies. We know when a human makes a mistake how to deal2273with it.2274    To the extent that AI makes human mistakes, we are covered,2275but AI makes different sorts of mistakes, so a lot of our2276systems for dealing with them just do not work. So, when I2277think about an AI being used as something important like a2278benefits administration, it is going to be how it is used. So,2279I am going to make this up. Like, we will have the AI do the2280work, but it is only allowed to say easy yeses. Cannot say no.2281Just say yes. Everything it does is reviewed, and all the noes,2282that anything that is hard, humans review. We can watch the AI2283get better or worse. We might give more responsibility, but2284again, with a lot of audit, it is really a matter of replacing2285versus augmenting.2286    And one of the early questions was about AI in procurement.2287So, I can think of two things where an AI is a negotiating2288assistant, assisting a human in negotiating. Walmart does this2289right now. Like, they use AIs because they have to buy a2290gazillion things, to help with that, but the humans are in2291charge, and also possibly using AI for auditing. There is a2292really interesting work done on using AI to figure out which2293slaughterhouses human ag inspectors should visit, right? There2294are a lot out there. We can pick at random, or we can have some2295algorithm. Again, the AI is not making the decision, the human2296is, so really is really thinking very carefully about how the2297AI and human work together where we can get the benefits of the2298AI without losing control. I mean, Mr. Perry talked about that2299is really important, but still getting the benefits of the2300speed or the scale that AI provides.2301    Ms. Randall. Thank you. You know, I think that that is a2302key difference that many of my neighbors and constituents are2303talking about, use technological tools for more information,2304for data gathering, for the ability to distill things, but when2305we are talking about, say, the life and death of individuals2306who depend on NIH-funded research or, you know, decisions that2307have to do with termination of individuals because maybe they2308wrote an email that says a word that people do not like in it,2309you know, this becomes really dangerous and has real human2310impact. So, putting in safeguards so that DOGE does not2311continue to make mistakes that are unable to be quickly2312corrected and that have real long-term impacts, I think, is2313very important for us as an Oversight Committee to keep on top2314of. Thank you, Madam Chair.2315    Ms. Mace. Thank you, and I will recognize Mr. Burlison for23165 minutes.2317    Mr. Burlison. Thank you, Madam Chair, for holding this very2318critical hearing on the Federal Government in the age of2319artificial intelligence. As we explore the AI's ability to2320transform this economy and the potential that it has to enhance2321Federal program efficiency, combat fraud, and deliver better2322services to the American people, we must address a fundamental2323challenge to this, and that is our Federal procurement system.2324    Our current system is riddled with barriers, lengthy2325certification processes, complex reseller networks, and2326multiyear contract vehicles that prevent agencies from2327accessing innovations or AI solutions, particularly from2328smaller companies. This procurement challenge not only hinders2329efficiency, but also threatens America's global leadership in2330AI.2331    In the last Congress, the 118th Congress, my bill, the FIT2332Procurement Act, was reported favorably out of this Committee2333and then successfully passed to the House with bipartisan2334support last year. I will be introducing the FIT Procurement2335Act again shortly, but let me point out that my bill directly2336tackles our outdated Federal procurement system. This act2337streamlines Federal acquisition to empower agencies to further2338acquire cutting-edge commercial technologies like AI. It does2339this by simply raising the simplified acquisition threshold2340from $250,000 to $500,000, increasing the micro purchase2341threshold from $10k to $25k, and allowing advanced payments for2342subscription-based software, which is what we see so readily2343today that the Federal Government is not able to take advantage2344of. With unanimous bipartisan support in the last Congress,2345this act is a proven commonsense reform to modernize government2346save taxpayer dollars and reposition the United States as a2347continued leader in the AI-driven future, and I look forward to2348hearing from our witnesses about that. So, let me get begin.2349    Ms. Miller, in your testimony you discuss why agencies2350should buy commercial technology rather than build custom code.2351As someone who was an in-house custom coder, I could not agree2352with you more. Could you follow up and elaborate on that?2353    Ms. Miller. Yes, sure. Well, the point I was making in my2354testimony about that was that you create technical debt when2355you are creating your code yourself, and because AI is moving2356so quickly, frequently it will be obsolete soon after it is2357built, and so it makes sense to contract with vendors who have2358more updated tools. I would just add that the bill that you are2359that you are referring to, given my own small companies'2360challenges, getting access to small pilot projects and proofs2361of concept, given those very low thresholds, this would be very2362beneficial to small innovative technology companies.2363    Mr. Burlison. Thank you. I do not have to now ask my2364follow-up questions. Is it true that AI can be used to detect2365fraud in the grant process effectively?2366    Ms. Miller. Yes, there are so many great use cases for AI2367in the fraud space when it comes to grants. We can scan through2368information using open-source intelligence. AI models, large2369language models can sort that information incredibly quickly,2370within seconds, and spot anomalies humans would not be able to2371spot. A lot of times what is happening today are grant2372processors are pulling up Google and opening up different2373screens and looking for things on different sites. Sometimes2374they have access to open-source intelligence tools, but they2375are still manually checking for that. AI makes that incredibly2376quick.2377    Mr. Burlison. Thank you. Mr. Shah, you have highlighted the2378Federal Government's procurement problem in their process.2379Based on your direct experience, do you believe that changes to2380the procurement process, allowing agencies to pilot products2381like yours before going enterprise wide, would allow smaller2382companies like you to better compete for contracts?2383    Mr. Shah. Yes, we believe that strongly. I think there are2384a variety of ways to do that: one, creating more POC-type2385programs. You know, a good example is the, I think it is called2386the investment horizon pilot program that the Office of the CIO2387of the Navy has implemented. But the challenge is, every agency2388seems to have their own set of ways that they procure new2389technology, so to the extent that we can help to normalize that2390will help us also navigate. You know, for example, through the2391FedRAMP process that I have described, of the $8.5 million,2392$600,000 I spent on consultants to help me figure out the2393process.2394    Mr. Burlison. Just to get certified, the program.2395    Mr. Shah. Yes. So, it is like, you know, I could have spent2396that on maybe more innovation or other areas. So, I think that2397if you have innovator programs or ways for us to start with2398other agencies that are less security-focused will give the2399help.2400    Mr. Burlison. Is it safe to say the Federal Government is a2401pain-in-the-ass client?2402    Mr. Shah. It is a complex web, and if you ever want to do a2403secret shopper with me 1 day and you can sort of watch how this2404plays out, I would be very happy to show you what it looks2405like.2406    Ms. Mace. Thank you.2407    Mr. Burlison. I yield back.2408    Ms. Mace. All right. I will now recognize Mr. Garcia for 52409minutes.2410    Ms. Garcia. Thank you, Madam Chairwoman. Thank you to our2411witnesses for being here today.2412    I want to just start by noting that we know that certainly2413AI can be an important tool for our Federal Government, for our2414Federal workforce, that we always want to make sure that we are2415centering those that are working in our government and uplift2416the work that our Federal workers do every single day. But we2417know that AI can navigate and help government become more2418efficient and intuitive in many cases. I think we should be2419honest about that. It can also help constituents get services2420faster. I have seen this happen at the local level. I have seen2421it happen, of course, and then happening at the state level and2422here in Congress as well. We also know that many agencies just2423need to move more quickly, be more efficient, and AI can be a2424tool when you respect the Federal workforce and work with the2425Federal workforce on how we make these implementations happen.2426AI can also be used, of course, on issues around red tape. I2427saw this happen, seeing happening in cities across the United2428States, and we can analyze data better and faster. We can also2429work to empower Federal workforce in ways that they can use AI2430to help them do their jobs, and I think that is an important2431piece of this.2432    It is also clear, though, that AI can be incredibly2433dangerous, incredibly disruptive, and certainly without2434guardrails can cause real harm to the American public and to2435the work that we are all trying to do. Now, deployment, of2436course, it is going to take investments in work, but I do have2437some serious concerns about how AI is used, and it is2438particularly in this Administration. And I want to talk about2439one of them, which I think is actually really, really2440important. So, this, of course, is Robert Kennedy, Jr., someone2441who I consider to be an extreme anti-vaxxer. I believe for him2442to be a conspiracy theorist, and certainly has no business2443being the Secretary of Health and Human Services.2444    Now, take one step back. We just went through a pandemic2445where 1.3 million Americans lost their lives, where businesses2446were shut down across the country, where healthcare was brought2447to the forefront of the public consciousness. And we know how2448important getting real medical vaccine testing information is2449to the American public, and right now we are dealing with2450measles outbreaks in Texas and other places. Information needs2451to be peer reviewed, fact checked, and done in a way that is2452responsible. But here, of course we know that there was2453recently a Make America Healthy report that was put out by RFK,2454Jr. in HHS, and I am going to include in the article just this2455quote: ``There were dozens of errors, including broken links,2456wrong issue numbers and missing or incorrect authors. Some2457studies were misstated to back up the report's conclusions, or,2458more damningly, did not exist at all. At least seven of the2459cited sources were entirely fictitious.'' What we do know from2460many folks that have read this report is it appears that AI2461played a role in developing a report about the public health of2462Americans. This is something that we should be incredibly2463concerned with.2464    And let us be clear. RFK, Jr. has a long history already of2465dangerous, unscientific beliefs. He said 5G and Wi-Fi cause2466brain damage. He said that you do not believe that HIV causes2467AIDS. He said that water can make you transgender. These are2468very concerning positions, and the fact that he puts out this2469report, and HHS, which has been long respected as an important2470Agency, puts out a report that is full of AI errors and AI-2471generated content should concern every Member of Congress on2472both sides of the aisle. This is incredibly devastating when2473you, on top of that, look at all the cuts that he is making to2474NIH and other health agencies across the United States, and so2475this is, for me, very concerning, and it should be for all of2476us.2477    Professor Schneier, is it fair to say that these kind of2478errors in important government documents and health reports2479that could be due to AI are concerning to us, and we should be2480concerned about?2481    Mr. Schneier. They are concerning, but put the blame where2482it belongs. Maybe the AI is not suitable for the task, or maybe2483the human who used the AI did not, like, check the work, and2484that would have been true if the human tasked an intern. I2485mean, so, it is really a matter of what is the process by2486which, whatever gives you your first draft, do you look at it2487and make sure it is correct? We see the same problem with2488attorneys submitting briefs into courts that have fake2489precedence. I mean, yes, the AI made the mistake, but it is the2490human who puts their name to it, who says this is correct. They2491are the ones responsible, and I teach this stuff. I tell my2492students to use AI, but you are responsible for plagiarism. You2493are responsible for accuracy.2494    Ms. Garcia. I agree with you completely, and, sir, I also2495spent 10 years in the classroom teaching as well, and----2496    Mr. Schneier. It is really weird now.2497    Ms. Garcia. And so, same, and I think that the idea that we2498are now seeing so much AI-generated content produced in our2499universities is a different topic, but let me just conclude2500with this. I think your point is exactly right. AI can be a2501tool, but there has to be the human element. The workforce has2502to be a part of it. It has to be a responsible user of AI and2503so that we do not end up with medical reports that are false.2504And with that, I yield back. Thank you.2505    Ms. Mace. Thank you. I will now recognize Mr. Grothman for25065 minutes.2507    Mr. Grothman. Thank you. I will start with Ms. Miller. In2508your testimony, you noted that the government loses more than a2509half a million dollars to fraud every year. Given that2510astonishing amount, what can the Federal Government learn from2511the private sector because we know the private sector would not2512put up with that, right?2513    Ms. Miller. Yes, it is a half a trillion, and, yes, it is2514an astounding amount. What we can learn from the private sector2515is, and I frequently say this when people ask me this question.2516You know, a bank invests in fraud prevention technology because2517they know that they are going to save more money with that2518investment than they are going to spend on the tool. A private2519sector company would never spend money on something it was not2520going to have a better return on investment, but in the2521government, because fraud is deceptive and we do not see it, we2522do not invest in anything to try to stop it. And so, it is this2523sort of happy sort of story that agency leaders can tell them2524because if they do not go looking for fraud, they do not find2525it, and then they do not have to invest in technology to2526prevent it. And so, then you have GAO coming out and saying,2527hey, we are actually losing half a trillion a year or more--I2528believe it is more--to fraud, and the agencies are not being2529held accountable.2530    Mr. Grothman. I will give you a kind of related question.2531Obviously, the answer is, you know, we are going to build more2532technology, and blah, blah, blah. We learned from DOGE a little2533bit about the work ethic of some of the Federal employees. As2534we spend more money somewhere to stop this fraud, do you think2535it should better be spent on hiring new government employees or2536maybe hiring out to somebody else who can try to detect this2537fraud?2538    Ms. Miller. Well, I do not think that government employees2539themselves have the tools to be able to find this fraud, so2540whether you hired more government employees or whether you2541contracted it out, I still think that there is such a lack of2542focus on it. I do think that concerted investment, whether that2543is in people or technology or both, will be needed if we really2544want to move the needle on this.2545    Mr. Grothman. OK. Do you feel, though, that the bulk of the2546work is going to have to be done by contracting out to somebody2547or other?2548    Ms. Miller. Do I? Well, I mean, I think, generally2549speaking, the Federal workforce does not have the skill sets2550that they need to be able to find fraud, and we should be2551skilling them up or we should be contracting it out, yes.2552    Mr. Grothman. OK. What is preventing the Federal Government2553from implementing similar AI practices that they have in the2554private sector?2555    Ms. Miller. A willingness to pretend that it is not2556happening because agencies are more focused on getting money2557out the door, and that is the goal, get the money out the door2558quickly and then take a victory lap because you spent the2559money, and not recognizing that there are people getting that2560money who are, you know, foreign state actors and criminal2561rings. But again, because it is underground, because it is not2562seen, we do not talk about it. We talk about it in these2563hearings, but we rarely talk about it, and the public is not2564aware.2565    Mr. Grothman. Yes, I guess part of the lesson should be,2566the less the government does, the better because they will2567always screw it up, right? That is what one of the lessons2568should be. Now we go to Mr. Thierer. On the first day of2569office, President Trump rescinded Biden's executive order on2570AI. Could you describe some of the damaging aspects of2571President Biden's executive order?2572    Mr. Thierer. Well, unfortunately, that executive order2573sought to do a lot of things that Congress should have been2574doing, Congressman, and really did not wait for Congress to2575take the lead on, on the questions of AI governance, more2576broadly. Some of the best practices that were there for2577agencies to follow, and some of the things that were2578recommended in terms of AI officers were fine. And Mr.2579President Trump has continued those practices in his executive2580order that replaced the Biden executive order. But I think,2581generally speaking, at a high level, I think the Biden2582Administration executive order was just very fear-based and2583treated AI as more like a curse to be avoided than an2584opportunity to be embraced.2585    Mr. Grothman. OK. In your testimony, you cited a study that2586stated utilizing AI systems would allow the Federal Government2587to unleash a productivity windfall for the U.S. Government2588worth up to $532 billion annually; not $531 billion, not $5332589billion--$532 billion. Can you describe what kind of2590productivity gains AI could unlock in the Federal Government2591and where the big savings are going to come?2592    Mr. Thierer. Yes, they are really twofold. As that report2593was noting, there are very specific types of targeted reform2594and oversight efforts that AI can help with to improve2595paperwork reduction, procedural issues, better administration2596of government. And then there are more broad-based productivity2597enhancements associated with AI itself and its use not only in2598the public but private sector that can, ultimately, benefit2599government administration.2600    Mr. Grothman. Could you tell me which agencies you think2601would most benefit? I do not know how you came up with $5322602billion, but is there any big agency? You grab $200 billion?2603    Mr. Thierer. Yes.2604    Mr. Grothman. What are the big----2605    Mr. Thierer. Yes. Well, a lot of Federal agencies could2606benefit from more efficient operations. Obviously, Department2607of Defense is one of them, but right now, important things2608being done at State and DHS and others just approve paperwork2609on things like immigration issues or paperwork issues, FOIA2610requests and so on.2611    Mr. Grothman. Thank you. Very helpful.2612    Ms. Mace. Thank you. I will now recognize Mr. Khanna for 52613minutes.2614    Mr. Khanna. Thank you, Madam Chair. I appreciate the2615testimony. Mr. Shah, of course you have been such a successful2616entrepreneur and leader in Silicon Valley. I appreciate your2617being here, so I will ask my first question to you. How do you2618think AI can help, particularly government employees, do better2619and be more effective in their work?2620    Mr. Shah. Yes, thank you. You know, the things that we are2621seeing in the private sector today with large organizations2622like Honeywell and like, you know, Northrop Grumman and the2623likes, they are benefiting from AI by making their employees be2624able to self-serve, get information that they are looking for2625and take actions across a variety of systems. You know, on any2626given day, an employee might have to swivel their chair 172627times to use different tools, and that becomes a very wasteful2628portion of their day, sometimes many hours. And so, to the2629extent that we can start to do that for Federal workers, for2630all folks who work for this government, I think we are going to2631see some benefits, benefits that will allow them to2632troubleshoot, to have better IT, benefits that will accrue to2633them from an H.R. perspective in terms of getting the2634information they need, making the changes, the benefit2635selections that go into that and also be able to work across2636different systems without having to be an expert.2637    I think one of the things that you all have pointed out,2638there are hundreds of enterprise systems inside the government,2639and you may only touch it once a year, twice a year. And so, if2640you can use AI as the overlay, kind of that front door, what2641you will see is your staff and all the members of the Federal2642Government be able to navigate these systems, which can be very2643complex, very easily using the most intuitive interface in the2644world, which is which is language and conversation. So, when it2645comes to that, you know, we have seen companies and our, sort2646of, world of private sector, be able to reclaim--a top five2647pharmaceutical was able to reclaim 75,000 hours of employee2648productivity by deploying this kind of AI across 44 languages.2649We have had a Fortune 500 aerospace manufacturer reclaim 752650percent of its support budget to be able to bring in2651technologies and leverage. I think that you are going to see2652both happier employees and better outcomes while also making2653the government more efficient.2654    Mr. Khanna. Appreciate that, and this is for anyone on the2655panel. How would you address people's concerns about AI2656displacing jobs, being something that will lead to layoffs or2657people not being able to work?2658    Mr. Schneier. I am happy to start. I think it is going to2659happen, and it is not going to be the jobs as we think. It is2660going to be, like, low-level lawyers, and that is going to be2661hard, I mean, because when you think about it, a lot of these2662professions are apprentice professions, and where do the senior2663partners come from if low-level lawyers or low-level doctors2664are being replaced by AI. Something society has to think about,2665whether there will be new jobs. In all previous revolutions,2666there have been new jobs to replace old jobs. This might be2667different. That is why you see a lot of talk about AGI, sorry,2668about universal basic income, UBI. Like, is this going to be2669different where we need to figure out how to give people the2670ability to survive, and we do not know. We do not know a lot of2671these things. So, those are very real fears, and I think we as2672society should think about them before they happen. We are not2673good at that, but we should.2674    Mr. Thierer. Congressman, just briefly, I would just point2675out that there have been a lot of government reports on this2676issue and a lot of efforts about government retraining, and so2677on and so forth. It is very hard to have a crystal ball to2678predict the jobs and skills that are needed in the future, but2679I will just point out that just 10 years ago, everyone was2680predicting a whole bunch of jobs that would be dislocated. The2681most famous one, and No. 1, was going to be radiologists. And2682just a couple weeks ago, the New York Times ran a piece to2683basically say there are more radiologists than ever, and they2684are utilizing AI to improve their jobs and free up time to do2685better things. And that is the story of technological change in2686sector after sector, is that we sometimes do not understand how2687complex human machine interactions work and actually can change2688jobs for the better and improve the employment output over2689time.2690    Mr. Khanna. I agree with being a technology optimist. I2691will say, though, that the unemployment rate for people between2692the ages of 21 and 29 these days with a college degree is 152693percent. And so, the challenge, I think, is, as Mr. Schneier2694pointed out, is how, especially at the entry level, for people2695particularly with college degrees, how are we going to create2696these jobs and how are they going to be able to use the tools2697of AI to be effective? But I appreciate all of your thinking on2698this and appreciate your being here for your testimony. Thank2699you.2700    Ms. Mace. Thank you. I will now recognize Mr. Donalds for 52701minutes.2702    Mr. Donalds. Thank you, Madam Chair. I think it is actually2703a good jumping off point from my colleague's comments about2704potential concerns of displacement. I think that, you know, as2705we looked at the evolutions of our economy over the last 1502706years, we have always hit these inflection points when there2707was a disruptive new technology or process for how large swaths2708of our economy operated, and there is always a concern about2709displacement. I think the difference now is that seeing the2710evolutions of AI, which, quite frankly, are just starting, and2711there are going to be things we are going to be talking about2712in a decade we would not even be thinking about today in his2713hearing.2714    I think it is more about making sure that kids going2715through the K-12 system primarily, but then also through higher2716education, are learning how to leverage AI and these tools to2717actually become far more productive, far more innovative, and2718then as a byproduct, having their labor be able to have a2719different set of value in the economy to come. And I think that2720it is an interesting thing, human beings and human psychology,2721we always try to find a way to have purpose regardless of2722whatever the dynamics are that are operating on the field of2723play of our economy. One of the largest data center operators2724in my home state of Florida, Equinix, was the first data center2725to partner with an SMR provider, Oklo. These nuclear, small,2726modular reactors can generate at least 500 megawatts of power2727to our energy grid and to power data centers.2728    Mr. Bajraktari. I think I got that right. That is not2729right. Close. My apologies. Can you share some ideas on how2730Congress and the Trump Administration can accelerate the safe2731deployment of clean, advanced nuclear energy and AI data2732centers to support our growing energy and computing demands?2733    Mr. Bajraktari. Thank you so much for your question. I2734think in the last couple of months, we have seen a really clear2735linkage between, you know, advancing AI and AI models and2736reliance on new modes of energy. I think my organization really2737is focused on the development of fusion energy, for example,2738because I think one of the pathways that now is showing real2739great success in our country is the emergence of private sector2740companies that are showing incredible successful models of how2741to get to fusion energy. We also have the national labs that2742have made major breakthroughs. You know, the Lawrence Livermore2743National Ignition Facility in 2022 have made the first fusion2744breakthrough. So, I think, in particular to your question,2745fusion is one of the pathways that I think can enable the2746energy demand that we will face going forward when it comes to2747AI models, but that is not going to be enough alone. So, I2748think we should pursue multiple pathways because we want to be2749energy independent and we want to have the best models2750available to the world.2751    Mr. Donalds. Mr. Thierer? Thierer? Thierer? I am not having2752a good day. Mr. Thierer, with the establishment of the special2753advisor for AI and crypto, what are some of the recommendations2754or priorities related to artificial intelligence that you would2755like to see them pursue?2756    Mr. Thierer. Well, one of them is what you just asked2757about, Congressman, which is making sure that America has2758energy independence and a greater diversity of energy sources2759to power the algorithmic revolution, and this is something the2760Trump Administration has taken steps to address with executive2761orders and other things like that. That is crucial. Obviously,2762a crucial part will also be the development of a national2763framework for artificial intelligence policy, just as we had2764for the internet before. We need to address this conflicting2765patchwork of policies that is developing. That is something the2766Administration and this Congress is obviously concerned with.2767And then there are a variety of other things having to do with2768national security matters and other investments in various2769types of algorithmic and robotic systems that are really,2770really crucial, among many, many other things that Congress and2771Administration are trying to do.2772    Mr. Donalds. Last point, I will just throw this out to the2773panel. I was sitting with my colleague, Mr. Perry. We were just2774kind of theorizing about what type of advantages AI2775advancements as well as, you know, I guess you could say2776quantum computing, what that could actually do to modernize and2777streamline governmental operations, not just Federal, but state2778and local. Just in your own expert opinions, what do you think2779the possibilities are in terms of how it would essentially help2780to redesign, streamline, and bring greater amounts of2781efficiency into the Federal Government? Like, where do you2782think is a good place to start, I guess I would say.2783    Mr. Shah. Yes, I would say that, you know, if you think2784about this new technology of Agentic AI, it allows for a new2785kind of automation. We have been building automation for2786several decades, but a lot of it was very brittle, and if2787something slightly changed under the hood, all of a sudden, you2788know, that automation would no longer work. With Agentic AI,2789you can actually overcome a lot of that by having these very,2790you know, smart, intelligent reasoners and systems that can2791then go and do the levels of automation that we have not been2792able to do before. And I think what that is going to do is2793really elevate all of us to be able to work on less mundane2794things and more sort of advanced thinking and creative ideas.2795    Mr. Donalds. All right. I am out of time, everybody. Thanks2796so much. I yield back.2797    Ms. Mace. Thank you. I will now recognize Ms. Crockett for27985 minutes.2799    Ms. Crockett. Thank you so much, Madam Chair. You know it2800is interesting because I do not know what all was taking place2801thus far in this hearing, but I do recall when I left, we were2802talking about Elon, and so I want to make sure that we do not2803forget him. I know that some people want us to forget Elon and2804act like he never existed, but we are not going to do that.2805    We are talking about AI today, and I can tell you what real2806people in my district care about. I had a few town halls and I2807talked to some seniors, and they were concerned about their2808data and their information, and how it was being used. And they2809wanted me to answer questions that I just did not have the2810answers to about what has Elon done with their tax information,2811their Social Security information, their health information,2812banking information, information about their credit histories.2813And instead of demanding answers, Republicans just blocked our2814motion to subpoena him to testify before this Committee. And I2815would ask at this point for unanimous consent to enter into the2816record a picture of who all was in the room when we took the2817vote, Madam Chair.2818    Ms. Mace. No objection.2819    Ms. Crockett. As well as since our clerk was ready and she2820knew who was in the room, the actual record of who was in the2821room at that time.2822    Ms. Mace. No objection.2823    Ms. Crockett. Thank you so much. Five months into the 119th2824Congress, and no one from the Administration or DOGE has2825appeared before this Committee to explain why they are invading2826the privacy of Americans. And the irony of it all is during a28272023 hearing, I remember Chairman Comer stating that, ``We have2828witnessed Big Tech autocrats wield their unchecked power to2829promote their preferred political opinions,'' and now he and2830Committee Republicans are here helping the world's richest man2831to do just that. They are letting him spy on companies2832competing for government grants, but worse, they are letting2833him spy on you, the American people.2834    Look, we can all agree that incorporating emerging2835technologies like artificial intelligence into government2836operations can be effective. However, these technologies should2837not be used to weaponize the government against the public. So,2838I am going to ask Mr. Schneier really quickly. Is it true that2839when it comes to any technologies, AI or otherwise, that they2840can be a double-edged sword?2841    Mr. Schneier. I think of technology as power, and the more2842the technology enhances your power, and whether your power is2843for good or for bad, depends on you, depends on how you are2844perceived. So, yes, technology has affordances. You know, we2845know that modern technologies have kind of built for2846surveillance, make surveillance easy. I mean, this is the most2847complex, effective surveillance device we have ever invented,2848and we all willingly put it in our pockets every morning. So,2849yes, technologies can be used for both good and evil, but think2850of the people behind it as the wielders.2851    Ms. Crockett. As the wielders. Yes, I absolutely agree, and2852I appreciate your answer. That is my concern. My concern is2853that these technologies will be used to weaponize the2854government against the public because everything that we have2855seen out of this particular Administration has been nothing2856short but weaponization. So, let us talk about some of the2857things that this Administration is doing.2858    They have announced they are going to use AI to scan the2859social media accounts of students who disagree with their2860policies. They have threatened nonprofits and universities--2861they are threatening their tax-exempt status. They have2862threatened journalists for reporting stories that are critical2863of the President. They are literally in court arguing to end2864birthright citizenship. And a lot of you all think only2865immigrants are going to be affected by this, but this2866Administration has already deported American citizens. In fact,2867in an interview with Time Magazine, Trump said, ``I would love2868to do that. We are looking into it.''2869    Madam Chair, I would ask unanimous consent to enter into2870the record an article published by ABC News titled, ``Trump2871Wants to Send U.S. Citizens to Foreign Prisons, Legal Experts2872Say.''2873    Ms. Mace. No objection.2874    Ms. Crockett. Thank you. It will not be long before more2875Americans are targets of this Administration. This2876Administration has refused to follow the law, and the2877Republicans are failing to do actual oversight. We do not know2878what they are going to do with their health records and Social2879Security information. In fact, we do not know what they are2880doing. I do not think anybody does, except we know that they2881are breaking the law. And it is frustrating to me because I am2882somebody that actually believes in science. I believe that we2883should be investing in our young people. I believe that we2884should not be taking research dollars away. I believe that we2885should be trying to come up with new technologies and that we2886should be leading the world. But unfortunately, I cannot trust2887these guys to do good things because, as it has already been2888said by Mr. Schneider, it is about who it is that is on the2889side that is asking for this to be done. And frankly, this2890feels like some bad movie that we are all in and that the2891villains are doing bad. So, with that, I will yield.2892    Ms. Mace. Thank you. I will now recognize Mr. Burchett for28935 minutes.2894    Mr. Burchett. Thank you, Chairlady. As the 435th most2895powerful Member of Congress, I have the honor of asking2896questions that have been asked ad nauseam. But when I ask them,2897answer as if this was the first time you said, and the proper2898response is, ``Congressman Burchett, that is possibly the most2899incredibly intelligent question that has been asked of me,'' so2900at least my clips will show that and the 12 people that are2901watching this thing are at home. They are probably not in my2902district anyway. They are probably in Marjorie's district.2903Anyway. Mr. Thierer? I get that name right?2904    Mr. Thierer. Right.2905    Mr. Burchett. All right. All right. The majority of East2906Tennesseans and, frankly, Tennesseans and Americans, are2907worried about their privacy being violated by government2908artificial intelligence programs. What do you think Congress2909can do to protect our privacy? And my biggest fear in all this2910is Congress. I mean, some of these cats up here still have2911eight-track tape players in their 1972 AMC Gremlins, and they2912are going to try to regulate AI. They cannot hardly spell2913``AI'' and now we are going to regulate it, so.2914    Mr. Thierer. Well, Congressman, thank you for that2915excellent question, and as someone who used to grow up in a2916Gremlin with an eight-track player, I feel great affinity for2917this question. So, I think that, obviously, we need to make2918sure that privacy is protected in the digital age, and this2919Congress has struggled with how to get that done. Obviously, we2920have had a multiyear, multi-decade debate about baseline2921privacy legislation, and a great deal of bipartisan consensus2922exists, that that should be a priority before you get to a lot2923of the other questions about how to regulate all the various2924specific concerns about AI. A lot of them go back to2925fundamental data privacy, data security issues. And so, I2926think, you know, let us not put the cart before the horse. Let2927us think about getting that done first. I know that is2928something Congress is turning to next.2929    Mr. Burchett. All right. Also regards to China, how do we2930maintain our dominance in the AI sector? Any of you all can2931answer that. I will ask Mr. Thierer first, if that is OK, on2932this.2933    Mr. Thierer. Well, a lot of it comes down to making sure2934that America diffuses first. We have wonderful products. We2935are, essentially, winning one part of the AI war but losing2936another. We are winning the part in terms of developing really2937sophisticated, high-powered frontier models and other systems2938and applications, but the reality is, is that China may be2939winning the diffusion side of things, and this is why it is2940very, very essential that we not miss out on making sure we2941have partnerships and deals, multilateral, mini-lateral2942approaches. The Administration has been trying to do this,2943obviously, recently, but this Congress should obviously make2944sure that we do everything we can to make sure American2945technology and products lead the world because it is not just2946about competitive advantage, it is about our values. Our values2947are embedded in our products, and, of course, the Chinese2948Communist Party values are embedded in their products of, you2949know, control, surveillance, and lack of human rights.2950    Mr. Bajraktari. Yes, sir. We got to do three things. No. 12951is we have to continue to out-innovate and out-maneuver China.2952Our platforms should be the dominant global platforms. No. 2 is2953we have to work with our allies and partners to spread our2954platforms to the global south and rest of the world. And three,2955we got to find ways to slow down China and make it impossible2956for them to deploy their capabilities and to use their2957capabilities against us.2958    Mr. Burchett. Can you give me that first one again? I can2959tell by your accent you are probably not from East Tennessee.2960    Mr. Bajraktari. I am not, sir.2961    Mr. Burchett. I am the only person in this room that is not2962speaking with an accent right now, so if you could say that2963first one again, brother.2964    Mr. Bajraktari. The first one, sir, was that we have to2965make sure that we out-innovate and out-maneuver China.2966    Mr. Burchett. Right.2967    Mr. Shah. I will just add, I agree with all the other2968panelists here and witnesses. I think that this is an infinite2969race, and the technology that is being discussed here today,2970AI, can essentially be wielded and constructed by any sort of2971group of smart people, and the planet grows smart people2972everywhere every day. And I think that to the extent that we2973can be the ones that others follow, that we can be the ones2974whose software, whose hardware, whose foundation models others2975use, we will have the sort of dominant leadership position of2976how this whole planet evolves.2977    Mr. Burchett. Ma'am?2978    Ms. Miller. Well, I would say that China is doing a very2979good job of figuring out how to defeat some of our security2980vulnerabilities. And I think that is one area that we have a2981ways to go is to protect the infrastructure because North Korea2982and China and others are getting better and better at2983infiltrating and exploiting our vulnerabilities.2984    Mr. Burchett. Sir?2985    Mr. Thierer. I am going to be contrary here. I think the2986U.S.-China arms race is an overblown metaphor. This is not the29871960s. This is not the cold war. Science happens globally. The2988transformer model invented by Google, it was published,2989everyone uses. DeepSeek, China comes up with this model. It is2990public. Everyone is innovating based on it. Ideas are flowing2991cross-border. The dominance is more companies than technology,2992so do not think about regulating the hardware, the software,2993the ideas. We cannot do that, and this is not the kind of world2994where that happens. It really is the companies and what they2995are doing. It could be U.S. companies, European companies,2996Chinese companies, other companies.2997    Mr. Burchett. All right. Thank you all. Chairlady, I yield2998back nothing.2999    Ms. Mace. OK. Thank you. I will now recognize Mr.3000Subramanyam for 5 minutes.3001    Mr. Subramanyam. Thank you, Madam Chair. I appreciate us3002having this hearing. I served in the second term of the Obama3003Administration as a technology advisor in the White House. I3004worked a lot with a lot of great groups who are dedicated to3005modernizing government, making it more efficient, bringing AI3006to government. And one of them was the U.S. Digital Service,3007which has now become the U.S. DOGE service. Another one was300818F, which has been completely set apart, completely gotten rid3009of it, and I will talk about them in a second. But one of the3010concerns I had about DOGE was that there is this notion coming3011in that they would come in and fire all these Federal civil3012servants and that AI would replace them, and that we would save3013trillions of dollars as a result, and so that is wrong for a3014bunch of reasons. One, even if you fire every Federal civil3015servant, you are actually not saving that much money in3016comparison to the deficit and the debt, but two, the Federal3017workers are so essential when it comes to helping utilize these3018technologies, and especially the technologists that we have3019hired in recent years.3020    And so, we have heard a lot about this already from people3021testifying and even from Members here, that it is not just3022about having AI, it is about having to deploy AI in a3023responsible way and in a manner in which we are not letting AI3024make mistakes. I will give you an example. In Michigan about,3025let us say, 10 years ago now, they deployed AI to try to find3026fraud, and they ended up finding a lot of fraud where there was3027not any fraud. And a lot of people got investigated or went3028bankrupt trying to defend themselves from getting sued, and3029then they realized that the AI had made a big mistake. And3030there are so many other examples of AI being deployed in an3031irresponsible way because we simply were not pairing them with3032a human element that could help sort of edit and deploy it3033responsibly.3034    And so, 18F was a great example of this. They were actually3035doing a lot of great work to modernize our Federal Government3036and bring AI to the Federal Government, and going agency by3037agency and helping us procure great tools to do that, and3038instead, they were all fired, basically. DOGE came in and fired3039the entire staff at 18F, and now we are trying to figure out3040how to deploy AI in government, but we are chasing away the3041very technologists who can come to the Federal Government and3042help us do that.3043    And so, if you want AI and modernization in government, we3044need the best and brightest to be coming to government, but3045instead, what we have is a brain drain on our Federal3046Government because we are trying, to quote the OMB Director,3047``put Federal workers in trauma.'' Who wants to come and work3048for the Federal Government if the people running the Federal3049Government wants to put you in trauma? Nobody. So, we are not3050going to attract the best and brightest technologists to help3051deploy AI, and we are certainly not going to have what we need3052to have pairing up with AI in order to deploy responsibly.3053    And then there are a lot of other things going on that I3054think are bipartisan, like, you know, making sure we update our3055legacy systems. I was just working with air traffic control,3056and I think their software is Windows 95. I have talked to many3057others in government who do technology, they still use COBOL,3058right? They still use really antiquated technology and coding.3059So, how are we supposed to deploy modern AI tools when we have3060these legacy systems, right? The other thing is, how are we3061supposed to recruit and retain great people to help transition3062and modernize our government when we are chasing them all away?3063    And finally, this AI human collaboration. It is already3064running into problems in the private sector, but it has to be3065even better in the Federal Government because like in that3066example with Michigan where people's lives were ruined because3067they deployed an AI technology irresponsibly, here, if you mess3068with an AI tool and it does not get people's checks on time for3069Social Security or Medicare or Medicaid, then people are going3070to go without healthcare. People are going to go without being3071able to pay their utility bills. It is going to have a terrible3072effect on the American people, and that is what this is really3073about.3074    It is not about just deploying AI so you can say you3075deployed AI. It is about making sure we are deploying3076technology for the benefit of the American people, and so I3077hope we can reverse course. I hope we can work in a bipartisan3078way to make sure that we are taking advantage of the advances3079in AI but doing so in a responsible way. I yield back.3080    Ms. Mace. Thank you, and I will now recognize myself for 53081minutes. I want to thank our participants and panelists and3082witnesses today for being here.3083    One of my concerns, coming from a technology background,3084like, I learned COBOL back in 1999, and a lot of our systems3085today, they are on legacy systems. We waste a lot of money on3086that, and then with the advent of AI and how fast technology is3087moving, you know, the Federal Government has real challenges3088with moving forward in a quick way and competing against our3089adversaries. So, this is my first question to all the panelists3090this afternoon, is if you could just do one thing that could3091advance the Federal Government in any way, any agency, any3092capability, if you could just do one thing today to help us3093utilize AI and technology to move the ball forward, what is3094that? What is the one thing you would do? Does not have to be3095something big. It could be a small part that makes a big3096difference. But what is that one thing you would do right now,3097if we could do it tomorrow, snap our fingers? What is it?3098    Mr. Bajraktari. Ma'am, I would say enable adoption, any3099models at any jobs in Federal Government, for memo writing all3100the way to execution, logistics, back office responsibilities,3101anything, but without enabling, you know, humans to start3102adopting these technologies, I do not think we will get to the3103next stage where we need to be as a country.3104    Ms. Mace. Mr. Shah?3105    Mr. Shah. Most of the innovation happening in AI is3106happening at small startups, and I think that, to the extent3107that you can look at a team of maybe 50 people and see how long3108it takes for them to get their product discovered and utilized3109by Federal Agencies, will determine how fast we can move, and3110to the extent that we can solve that, I think it would make a3111big difference.3112    Ms. Mace. Ms. Miller?3113    Ms. Miller. Pick a high-value process that can be automated3114using agents and deploy them. So, for an example, continuing3115disability review in Social Security, we have seen a 9 to 1 ROI3116on that process of reviewing continuing disabilities that is3117using people. That ROI would be at least probably doubled if we3118used AI agents, and we would save a lot of money in a very,3119very high-value area. So, pick a really high-value area, deploy3120agents, and measure the difference so that you can use that3121measurement to inform other investments.3122    Mr. Thierer. Chairwoman, you are going to like my answer3123because I am going to say we need the AI Training Extension Act3124to be moved.3125    Ms. Mace. I love that.3126    Mr. Thierer. We need better literacy and skills government-3127wide. We have a lot of smart people working in government, and3128sometimes it is just a matter of getting them moving in the3129right direction and the right training to get the job done.3130    Ms. Mace. Mr. Schneier?3131    Ms. Schneier. I want to see a model that was not built by a3132corporation. I want to see a public AI, that democratic3133processes are different, they have different ordinances, they3134have different requirements, and to get a model that will be3135trustworthy for the U.S. Government. I think it needs to be3136U.S. Government built. I do not think it can be corporate3137built. So, funding and research into a public AI will give us a3138different sort of AI. I think it is going to be interesting.3139    Ms. Mace. It is interesting. You know, we have all this3140technology at our fingertips, but, like, going back to Ms.3141Miller's comments about fraud and abuse, we have this3142technology, but it is almost like we write these checks and3143then we cross-check it later. So, there is a System Called3144DoNotPay. It is called continuous monitoring. It still gives3145agencies match results the next business day, for example, and3146in 2025 that is dial up speed, right, when you can have it3147within seconds. And taxpayers are footing the bill for a 24-3148hour lag when there are fintech companies that are out there3149that can flag fraud in milliseconds, like that technology3150exists. The March 20 executive order would force agencies to3151share all unclassified records with authorized officials, yet3152most agencies are still swapping spreadsheets as opposed to3153using APIs and technology that is, again, readily at their3154fingertips.3155    The Secretary of Labor now has statutory access to state3156unemployment insurance files, but those fees are not wired into3157DoNotPay as of today, right now, and it seems like we are still3158paying benefits first and then cross-checking later, which,3159again, makes it rampant for fraud that we have had year over3160year over year. And then the IRS individual master file is3161written in COBOL in the 1960s, and it still drives every tax3162refund. Like, I cannot even wrap my brain around. I mean, if I3163saw a line of COBOL right now, I would probably throw up, and3164then to know one of the most important agencies for American3165taxpayers, you can get your money back from the government, it3166was written in COBOL from the 1960s. So, how can an algorithm3167catch refund fraud in real time when the source of the data3168lives in a mainframe computer?3169    And then the last thing I want to say is that GAO says that3170the Federal Government blows 80 percent of $100 billion IT3171budget just keeping 1970s systems on life support. I would love3172to know after this, and I will send a question follow-up, but3173if there was one legacy IT system you could cancel today, what3174would it be? And I do not have time, but that is a follow-up3175question I would love to ask all of you. What is the one legacy3176IT system you would just cancel tomorrow, if you could, to make3177our systems better?3178    Ms. Mace. Thank you so much, and I yield back. I will now3179yield to Ms. Lee for 5 minutes.3180    Ms. Lee. Thank you, Madam Chair. As we talk about AI, I do3181not think we can afford to ignore what is going on right now in3182our Federal Government. So, the latest out, we know that Musk3183and the DOGE children went into every department and accessed3184their datasets. Initially, they were digging around and using3185AI to cut government funding, but it has become clear now that3186the goal has been to amass and merge these different pots of3187data into one data base.3188    Trump has now tapped tech billionaire Peter Thiel's3189Palantir to head up that master data base. We are not talking3190about just inter-government memos here in the data base. We are3191talking about Americans' Social Security numbers, home3192addresses, credit card numbers, medical diagnoses, and more.3193All that information would just be in one place, and then they3194are going to unleash opaque algorithms that no one understands3195but them. So, they will be able to do whatever they want, and3196we are supposed to just trust that they are working within our3197best interest. And we probably will not conduct much oversight3198because we have not been able to bring Elon Musk or anybody3199associated with DOGE into the Committee on Oversight and3200Government Accountability [sic].3201    So, Mr. Schneier, looking at the immediate risks of this,3202is there a concern that having all of this information in one3203place can make it easier to hack or steal?3204    Mr. Schneier. Well, that is the fear, that we keep this3205data separate for partially security reasons. I mean, there is3206a reason why that when data shared between agencies are done3207under certain rules and why there are security protections3208around the data. So, it is the data being consolidated that is3209a risk. Most of the data leaving the government data bases3210where they were, that is a risk. They are being moved into3211private companies. They are being used by algorithms, well, you3212said no one understands, but them. It is actually worse than3213that. No one understands, including them. These algorithms are3214opaque, and we do not know the quality of the data as data is3215combined.3216    Talking about COBOL, a lot of that is how the data is3217presented and making sure that it is the same across. In my3218initial statement, I talked about that story about 150-year-3219olds getting Social Security. That was a problem with3220understanding how the COBOL was not decoded.3221    Ms. Lee. Certainly.3222    Mr. Schneier. And if you do not know that, you are going to3223have bad inferences that no one is going to understand and no3224one is going to check.3225    Ms. Lee. Thank you. So, beyond the concerns with the data3226all being in one place, one less secure place, we also need to3227be concerned about with what Trump and these tech billionaires3228are going to do with that data. Just very quickly, and ``yes''3229or ``noes.'' Mr. Schneier, could an AI system theoretically be3230used to decide if someone qualifies for Medicaid?3231    Mr. Schneier. Certainly.3232    Ms. Lee. How about used to make decisions around veterans'3233benefits?3234    Mr. Schneier. Yes.3235    Ms. Lee. Could AI be used to make hiring decisions?3236    Mr. Schneier. It can make any yes or no question you want.3237    Ms. Lee. Thank you for entertaining that. Trump and his3238Administration can take all this collected data and use it to3239profile people, to make decisions about them, to deny them3240benefits, and we know from his Twitter rants that he is3241incapable of being unbiased. Then they are trying to prevent3242any state-passed guardrails on AI through the provision in the3243reconciliation bill that you guys unseriously called the Big3244Beautiful Bill. That provision would ban any state regulation3245of AI for 10 years.3246    Our government cannot just play loose and fast when it3247comes to people's personal data and lives. We have already seen3248the consequences of improperly deployed AI. Michigan had to3249refund $21 million to residents who were wrongfully accused of3250unemployment fraud from their AI system. A tenant screening3251system called SafeRent was accused of violating the Fair3252Housing Act because their AI algorithm disproportionately3253scored Black and Hispanic renters lower than White applicants.3254    Most of this technology is still so new, it needs time to3255be vetted and improved. Look at what happens when AI is used3256just in government reports. The Make America Healthy Again3257reports cited to multiple studies that did not exist. Can you3258imagine if ChatGPT was deciding if you qualify for housing or3259not? That is not to say that all AI is bad, but it is dangerous3260to work quickly. When you move fast and break things in the3261government, people get hurt.3262    Carnegie Mellon University, which is in my district, has a3263Center for AI Standards and Innovation, and has been working3264with the National Institute of Standards and Technology to make3265a methodical approach to assessing the benefits and risks of AI3266models. Thinking about the ethics of AI, work like theirs is3267vital to ensuring that AI programs are used safely, accurately,3268and without bias to the best of our ability, recognizing that a3269society without biases is far off. But that is the approach the3270Trump Administration should be taking, not packaging up all of3271our personal data and handing it over to unvetted tech3272billionaires. I thank you all for your time, and I yield back.3273    Ms. Greene. [Presiding.] The gentlelady yields. I now3274recognize myself for 5 minutes.3275    We just passed the One Big Beautiful Bill through the House3276of Representatives, and now it has gone on to the Senate where3277they will be making changes to the bill. If there are any3278changes made to the One Big Beautiful Bill, it comes back to3279the House for a vote and we get a second bite at the apple.3280    Now, while this bill was going through committees and being3281discussed in the House of Representatives, no one on either3282side of the aisle that I know of, Republican or Democrat,3283brought up this particular clause on one single page in an over32841,000-page bill, and I want to bring this up because I think AI3285is incredibly important. I support AI and many different3286faculties. However, I think that at this time, as our3287generation is very much responsible, not only here in Congress,3288but leaders in tech industry and leaders in states and all3289around the world have an incredible responsibility of the3290future and development, regulation and laws of AI.3291    It is such an important responsibility, it literally3292affects our children and grandchildren's future. So, I take3293this responsibility incredibly sincere, so much so that this3294past week I have come out in full transparency and said when I3295voted for the One Big Beautiful Bill, I did not know about this3296clause. I thought I was voting on taxes, energy, and border3297security, and this clause right here says a moratorium. In3298general, ``Except as provided in Paragraph 2, no state or3299political subdivision thereof may enforce during a 10-year3300period, beginning on the date of the enactment of this act, any3301law or regulation of that state or a political subdivision3302thereof, limiting, restricting. or otherwise regulating3303artificial intelligence models, artificial intelligence3304systems, or automated decision systems entered into interstate3305commerce.''3306    That is a comprehensive 10-year moratorium. Actually, what3307that is, is it is a pause for 10 years in federalism. This3308right here, if passed into law, will tell states they cannot3309regulate or make laws regarding AI. That means the Federal3310Government is the only governing body that can regulate or make3311laws for AI in the United States of America. And I would like3312to ask each of you, do you support federalism? Yes or no. I3313will start with you. Do you support federalism? Yes or no?3314State rights. It is a yes-or-no question.3315    Mr. Bajraktari. I do not think I am able to answer this3316question based on my background, to be honest with you.3317    Ms. Greene. OK. Then I take that as a no. Mr. Shah, do you3318support federalism?3319    Mr. Shah. I do support state rights, and just from a small3320company standpoint, I would say whatever we can do to make it3321easier for a young company to navigate the 50 states will be3322very important, because otherwise----3323    Ms. Green. Well, either you support federalism or you do3324not, one or the other. It is a yes or no.3325    Mr. Shah. Yes.3326    Mr. Greene. OK. Ms. Miller, do you?3327    Ms. Miller. Yes.3328    Ms. Greene. Yes, Mr. Thierer?3329    Mr. Thierer. Yes. In line with the White House and with3330Speaker Johnson, I agree that federalism is important, but3331federalism is a two-sided coin of both states' rights and3332interstate commerce that we need to protect.3333    Ms. Greene. The Constitution is clear on federalism. It is3334one side, state rights or none. Mr. Schneier?3335    Mr. Schneier. I think that provision is nutty.3336    Ms. Greene. OK. So, do you support federalism? Yes or no.3337    Mr. Scheiner. In general, yes.3338    Mr. Greene. Oh, in general, so that is vague. I am not3339sure.3340    Mr. Scheiner. It is vague, but that provision is nutty.3341    Ms. Greene. OK. So, I still did not get an answer clearly3342on federalism. Can you predict the future of AI in 1 year, 53343years, or 10 years? I am sorry. I apologize.3344    Mr. Bajraktari. Yes. No, no problem. No, no problem. No, I3345do not think anybody can predict that. I think every model has3346become bigger and better and more transformative. I think----3347    Ms. Greene. Right, so the answer is no. I am short on time.3348Mr. Shah, no or yes?3349    Mr. Shah. I think it is hard to predict 5 years.3350    Ms. Greene. Yes or no. Can you predict that period?3351    Mr. Shah. No.3352    Ms. Greene. OK. Ms. Miller, can you predict AI?3353    Ms. Miller. No.3354    Ms. Greene. Mr. Thierer?3355    Mr. Thierer. No.3356    Ms. Greene. No? Mr. Schneier?3357    Mr. Schneier. Not a chance.3358    Ms. Greene. No, you cannot. In the state of Georgia, jobs3359are extremely important, and AI is, whether we like it or not,3360and it is helpful in many ways, and it is not helpful, will3361replace jobs. I have a manufacturing district, and the3362manufacturing companies are important, but so are the people's3363jobs. And if our state cannot regulate or make laws to protect3364people's jobs, people are going to go hungry. They are not3365going to have paychecks. They are not going to put roofs over3366their families' heads. And I think this is such a serious3367issue, that we need to look at this through--I will state it3368very clearly, I am pro humanity. I am not pro transhumanity.3369And when it comes to AI and regulation, when we get to vote on3370this bill again, I will be voting no because of this clause,3371and we will be working on this further. Thank you so much for3372being here today. I now yield to Ms. Simon.3373    Ms. Simon. Thank you, Madam Chair. Thank you all for coming3374today and sitting in this very chilly room. I think we have all3375learned so much. I come from the Bay Area, which is, as you all3376know, the technological gold mine for the artificial3377intelligence community. I am so proud of that. In March,3378actually, I participated, I actually convened a table with AI3379innovators and leaders in my congressional district on how AI3380is and will continue to enable access to high-quality3381healthcare, data sharing, and, of course, opportunities for3382physicians and folks in the industry to expand access to the3383many wonderful folks who need it and deserve it.3384    We already know that AI is advancing lifesaving, early3385detection for cancer, Alzheimer's, heart disease, and I can go3386on and on, and can save clinicians time in charting and note-3387taking and better spent on patients and actual working with3388families and folks who need them. This technology helps3389patients self-manage their diseases with suggested resources3390and reminders to take medications. Again, as you all know, I3391can go on and on. And as somebody, myself, with a visual3392impairment, I was born legally blind. I cannot tell you how3393artificial intelligence has led the disabled community to be3394able to live with more dependence and more self-determination,3395and it has exceeded, in this moment, our wildest expectations.3396The guardrails are needed, absolutely regulation is needed, but3397again, so many of us from impacted communities are so excited3398about what this technology allows for us to do.3399    The Federal Government has this responsibility and duty not3400only to set a standard around data privacy and security, but to3401follow the best practices. I have a lot written, but time is3402short, so I just actually have a couple of questions that I3403will just go straight to, and again, I want to thank you all3404today, and I am sure all of our offices will be working with3405not only you all, but leaders in the industry moving forward3406forever. So, my question is actually for Mr. Schneier.3407    You know, I have met with, as I mentioned, AI companies in3408my district who manage very sensitive health data, very3409sensitive health information with great care and sensibility3410and sensitivity, and these are the companies that I have met3411with. They only use de-identified patient information, data3412sharing, and encryption, and otherwise control access. Can you3413describe how our current Administration's actions differ from3414the industry standards and best practices around data privacy3415and security?3416    Mr. Schneier. Oh, in so many ways. So, we have seen3417examples of DOGE dismantling security protections inside3418organizations. We have seen them taking data out of secure3419enclaves and taking it off premises, and we do not know where.3420We have seen them combining data in ways that are contrary to3421law. If you think about it, healthcare is controlled by HIPAA.3422There is a regulation about health care privacy. De-identified3423data is a science. The census has a whole program on how to de-3424identify data to be used by researchers without revealing3425private information. So, we can do this if we choose to. DOGE3426chose not to.3427    Ms. Simon. Well, I have a little bit more time. Continue.3428    Mr. Schneier. All right. So, I would want to add, moving it3429into AI models has its own risks, and it is several risks. It3430is the risks of the model being trained on data that is private3431and then could make it public. We see researchers being able to3432pull private data out of a model, so without safeguards, the3433training process fails in privacy. Then the data will be used3434as an input to whatever the model is to produce decisions, and3435that also can be a violation of privacy. So, all through these3436processes, we can build and we can build safeguards. I think3437that is your point.3438    Ms. Simon. Yes.3439    Mr. Schneier. We just have to do it, and there is a reason3440why you do this slowly and deliberately within organizations3441and not just pull the data out.3442    Ms. Simon. Thank you, sir, and actually, this is a question3443that, as you are talking, just derived in my mind. Anyone on3444the panel want to talk about the innovations in healthcare and3445AI, particularly around diagnostics? I only have about 303446seconds, but I think it is an exciting innovation and folks3447will live longer, I believe, from what I have heard. Anyone3448want to jump in?3449    Mr. Thierer. Yes. Congresswoman, thank you for your3450leadership on this front. I will submit for the record a series3451of reports that my organization has put together on AI and3452public health where we cite specific ways that AI is being3453utilized in drug discovery, medical devices, and so on and so3454forth. I will just mention that we have had a 50-year war going3455on cancer and we have not made as much progress as we need to,3456but last year, 608,000 people died of cancer, and a lot of3457these things can be addressed now with the power of artificial3458intelligence.3459    Ms. Simon. I appreciate that, and, Madam Chair, thank you3460for your time and allowing me this time. My husband died of3461cancer. I know that he could have lived longer if we had these3462technologies. We got 2 years with him. Again, that was 10 years3463ago. Changed our lives forever. Thank you for your work, your3464innovation, and clearly, we need safeguards, but the future is3465bright. Appreciate you all.3466    Ms. Greene. The gentlelady's time has expired. I now3467recognize Mr. Timmons from South Carolina.3468    Mr. Timmons. Thank you, Madam Chair. The greatest national3469security threat we face is debt. We have $37 trillion in debt.3470We have a $1.8 trillion annual deficit. The work this3471Administration has done over the last few months has been3472designed to try to find waste, fraud, and abuse to help3473mitigate that. They are also working to help grow the economy3474through tax reform, through renegotiating trade agreements,3475through deregulation, through streamlining permitting. The list3476goes on and on. But I think that AI has a huge opportunity to3477grow the work that DOGE has done as it relates to seeking out3478waste, fraud, and abuse.3479    So, the main thing that they did early on was to get all3480the government agencies to communicate using the same language,3481to code the same, to then be able to create transparency and3482track data from all the different government agencies. And that3483effort was really the beginning because now that everything3484goes through Treasury, now that we can track payment systems,3485we now have to layer on technology, and some of that is fairly3486simple.3487    For example, I have introduced the TABS Act, Timely3488Accurate Benefits, and that what that act does is it uses a3489pilot program that was done in Missouri where they, instead of3490using antiquated ways of confirming income eligibility for3491Medicaid, they use third-party website to verify eligibility3492immediately. And then the second you are no longer eligible,3493you are removed from whatever benefit it is, and so that3494resulted in a 17-percent reduction.3495    So, the way this technology works is you say, all right, I3496am on hard times. I want to get whatever government benefit,3497and they say, all right, well, you got to log into your bank3498account. And then when you log into your bank account, it says,3499OK, you have a Venmo, you have a Cash App, you have a Zelle,3500you have to log into that, too, and then immediately they say3501you are eligible or you are not eligible, and so you are3502actually getting benefits faster than normal. And this has a3503huge potential to save money and that 17 percent of Medicaid is3504about $150 billion if you do it over the entire Federal3505Government.3506    And, you know, one thing that the One Big Beautiful Bill3507did was instead of confirming eligibility once a year, which I3508think is ridiculous, we are now confirming eligibility twice a3509year. Well, why don't we confirm eligibility every day using3510technology? This is not hard. The technology exists. All we3511have to do is implement it. So, that is one way that we can3512further find waste, fraud, and abuse because I think we can all3513agree that if you are not eligible because you make too much3514money or you are not eligible because, and this is the next3515step, because you are not the person who you say you are, then3516you should not get government benefits.3517    So, I am working on TABS 2, which is using technology to3518confirm you are who you say you are, and this is just using3519billions of publicly available pieces of data that, you know,3520for example, if you have an email address that was created in3521the last year and you are 45 years old, you are probably not3522who you say you are. If you have three different phone numbers,3523that is not normal. So, they are able to use technology,3524publicly available information, to confirm an individual3525requesting benefits is who they say they are. And honestly,3526that would not only help means-tested programs, it would help3527any government program. If you are saying you are who you say3528you are for Social Security, I mean, that is not means-tested,3529but we could use it to seek out waste, fraud, and abuse.3530    So, those are two things, two ways we can use technology to3531reserve government benefits for American citizens that deserve3532them. How can we grow that effort? How can we build on that3533effort using AI? Mr. Bajraktari?3534    Mr. Bajraktari. Thank you. Sir, I think those two examples3535indicate the potential that we have in AI adoption for3536government services. I think we have got to do three things.3537The underlying issue that we have here is that we need to3538upgrade or modernize our IT infrastructure because you cannot3539deploy AI if you do not have the necessary infrastructure3540underneath that, and so to do that, we need to do three things.3541No. 1 is we need to upgrade our platforms, No. 2 is we have to3542update our policies, obviously, and No. 3 is we have to work on3543retraining and re-skilling our people.3544    Mr. Timmons. Ms. Miller, what are your thoughts?3545    Ms. Miller. There is a lot we can do in this area. I mean,3546this is what the software company that I created just does3547this. And mining open-source intelligence to identify whether3548somebody is who they say they are, is the lowest hanging fruit3549and the best use cases we can do in the fraud space for3550government programs, and not just means-tested programs, every3551grant program. We put a trillion dollars in Federal money out3552every year to states for grants, and we have very little3553understanding of where that money is going, but we can use AI3554and open-source intelligence to very quickly identify if it is3555a shell company or even an organized crime ring that is3556actually getting that money.3557    Mr. Timmons. But on the organized crime front, a lot of3558these involve changing addresses once they request benefits and3559things of that nature, and AI could easily say, oh, it is3560weird. There is a P.O. box in Eastern Europe that gets 433561checks a day. That is weird.3562    Ms. Miller. A hundred percent. There is geolocation. There3563are all kinds of behavioral biometrics that AI can flag. You3564can use, you know, probably dozens of different data inputs to3565identify fraud today using AI.3566    Mr. Timmons. And this is not a partisan issue. I think3567everyone can agree that only people that deserve benefits, that3568are entitled to benefits, should get them. So, again, I think3569using technology to make sure that we are being wise with our3570taxpayer dollars is something that is a no-brainer, and I look3571forward to working with my colleagues across the aisle to do3572just that. Thank you, Madam Chair. I yield back.3573    Ms. Greene. The gentleman yields. I now recognize Mr. Min3574from California.3575    Mr. Min. Thank you, Madam Chair. I appreciate the3576opportunity to discuss artificial intelligence in the Federal3577Government. Of course, AI has been around for a while, but it3578is recently hit, I think we all agree, a tipping point of3579exponential and profound growth. Innovation and advancement are3580now being measured in days or even seconds or minutes as3581opposed to years, and this rapid development of AI is poised to3582radically transform our society and our economy. And that is3583going to create a lot of great benefits, I believe, but also a3584lot of challenges, including to our workforce, our national3585security, data privacy, and a whole host of other issues, and3586these benefits and challenges of AI are also important to3587consider as we contemplate bringing more AI into our Federal3588Government.3589    And one important point, and I know it has been made, but I3590just want to reemphasize this, AI is only as good as the inputs3591it receives, inputs it is trained on, the people who train it,3592and the objectives is trained to try to achieve, and that is3593something that I do think is being lost right now. In the rush3594to think about AI, I think we have to think about what some of3595the goals are. And a lot of mention has been made of the rash3596and reckless actions made by the Trump Administration,3597specifically Elon Musk and DOGE, around AI, and again, I think3598what are we looking at as far as goals? I think we want to make3599sure that the missions of our different agencies, of the tasks3600they are meant to accomplish, is what is in mind, and not just3601simply cutting for the sake of cutting.3602    And so, for example, you know, when we think about AI, DOGE3603may be using AI right now to surveil communications at Federal3604agencies, performing potential cuts, including staff and3605programs. They have stated that they want to use generative AI3606to automate much of the work previously done by civil servants,3607but there are a lot of potential mistakes there, as well as the3608data privacy aspects that have been touched on in previous3609questions. But this data privacy, and we do not actually know3610at this point what Elon Musk and DOGE have been able to scrape.3611They certainly have had access to a lot of our data, illegally3612so I would add, and what have they done with that? We have been3613told that they have downloaded them onto private servers. This3614Committee has not done any oversight on this question, but I3615think it is an important one. As we think about AI, is Grok or3616Elon Musk's other AI models, are they being trained on data3617that is only available because of his position at DOGE? This3618also creates a lot of national security risks as well.3619    And so, I guess my first question is to Mr. Schneier. From3620a national security perspective, what safeguards should we be3621thinking about implementing to make sure that we are protecting3622sensitive data while still promoting artificial intelligence,3623design, and innovation?3624    Mr. Schneier. I mean, we have the safeguards. The3625regulations are in place that protect the data. We know how to3626do it. We have been doing it for years, for decades, and it is3627a matter of implementing them. The problem, as you said, was3628what DOGE did was illegal. It was contrary to the laws, and3629when I think about using AI in organizations, it is a matter of3630bringing the technology in and using it in a way that makes3631sense. There is a lot of talk about using it to find fraud.3632That seems like a good use.3633    I would want to add, I see Mr. Timmons left, but how about3634tax fraud? That is, like, the for-profit arm of the government,3635and we get a lot of bang for our buck for looking for tax3636fraud. I think AI would be great at a first thing, but it is a3637matter of keeping the data in the organizations with the laws3638that protect it.3639    Mr. Min. And I think Isaac Asimov wrote a book about this a3640long time ago, but something similar that we ought to be3641training AI, as we think about it, to follow basic rules and3642principles that we want them to follow first and foremost3643instead of just thinking about the end goals in mind. Now,3644again, I want to get back to the point of goals being3645important. When we look at Social Security, when we look at3646programs like Medicare and Medicaid, it is easy to simply just3647cut costs. But if we are diminishing service, if we are just3648looking at cutting costs and not at the service being provided3649or the efficiency of the service being provided, we end up with3650huge mistakes. At the same time, if you have a bunch of coders3651out there who do not have any context about the program, how it3652is been administered, you end up with huge problems, as we saw.3653    You know, we saw that DOGE flew into the Administration3654without even understanding the basics of how Social Security3655was coding its different entries in the COBOL coding process.3656And so, in that review, they noticed entries that made it seem3657like the Administration was paying individuals benefits who3658were aged 150 or older. Trump and Musk actually claimed this3659falsely, and again, it was because you relied on computer3660wizard geniuses, like, the 22-year-old coder named Big Balls,3661instead of actual Administration professionals who understood3662what was going on. Similarly, with Medicare and Medicaid, the3663goal here should be to maximize healthcare in an efficient way,3664not simply to cut costs. The cheapest way to administer3665healthcare, of course, is just to let everyone die. That is3666cheaper than administering healthcare, and so we cannot lose3667the forest before the trees here. So, I see I am out of time.3668    The last point I just want to make is the 10-year3669moratorium on AI regulation, something that is outrageous,3670something that was slipped in, in the dark at night, it is a3671shocking provision, and it was shocking that my colleague,3672Madam Chair Taylor Greene, said she would vote against the3673entire Big Beautiful Bill if this provision was not set out.3674    Ms. Greene. The gentleman's time has expired. I now3675recognize Mrs. Luna of Florida.3676    Mr. Min. Thank you.3677    Mrs. Luna. Thank you, Madam Chair. You know, I want to3678thank you all for being here today. One of the biggest concerns3679I have with this whole discussion on AI is a failure to admit,3680first and foremost, that it does appear that in the next 103681years, that the way that we have grown up and know life with AI3682will be forever changing. With that being said, there are a few3683differences that I think most people acknowledge and that AI3684lacks a few things, one being a soul, also empathy. And we are3685not gods or God, and so I think that we are playing a dangerous3686game, especially moving into a future that is simply unknown3687right now, especially if we do not essentially develop the3688first AI super weapon. Essentially, that is what this is.3689Whoever does will essentially control the world, and I think3690that that is a very serious topic that needs to be discussed3691right now.3692    But specifically, to transhumanism and coupling of AI with3693humanity, what regulatory frameworks can be established to3694ensure AI-driven transhumanist technologies, like brain-3695computer interfaces prioritize human safety and consent? I am3696opening this up to everyone on the panel. I have a few3697questions, so I would like to get through this, so just please3698limit your responses. Mr. B, if you want to go first, or down3699there, either/or.3700    Mr. Thierer. I will just briefly say, Congresswoman, that3701there has been a lot of work done on AI governance by multiple3702bodies and entities. I just want to cite something that3703actually came from the Biden Administration on this about the3704applicability of so many existing laws and regulations to AI.3705And their leading regulators, or four large agencies, said3706before leaving office that, basically, we have the ability to3707enforce the respective laws and regulations to promote3708responsible innovation and automated systems, and that,3709basically, AI does not get an exemption from civil rights laws,3710from consumer protections, unfair and deceptive practices,3711fraud, so on and so forth. So, we have a lot of policies that3712do regulate many of the fears you are raising.3713    Mrs. Luna. Is there the possibility of regulation, though,3714that exists that would enable transhumanist enhancements that3715exacerbate social inequalities or basically financial3716inequalities, creating an elite class of enhanced individuals?3717I bring that up because we are talking about low-level jobs3718potentially being removed from the workforce, people that have3719and can afford the ability of, for example, implementing or3720implanting chips, given access to unknown amounts of, you know,3721knowledge, and essentially creating the first superhuman, and3722this is a real concern. I know it might come across as funny to3723other people, but I am genuinely concerned about this.3724    You know, from a bipartisan perspective here, we are3725talking about humanity versus machine. Hopefully, in a flowery3726world, we would have it set up to where it could, you know,3727improve society for the best, but I am also in politics, and I3728have a very, unfortunately, sometimes negative perspective on3729the world because I have seen the worst of humanity in this3730job.3731    Mr. Schneier. So, science fiction speaks of this a lot. I3732mean, there are a lot of people who think about this notion of3733haves and have-nots, and how technology will exacerbate that,3734and that is going to depend on how it is available and how much3735money it costs and who can get it. Nita Farahany writes a lot3736about AI and brain interfaces and what we can do to protect3737humanity in that world, so I do recommend talking to her. She3738is the smartest person I know on that particular question.3739    Mrs. Luna. OK. And also, what role can AI itself play in3740monitoring and mitigating risks associated with transhumanist3741technologies, such as detecting biases and enhancement3742distribution?3743    Mr. Shah. I think a lot of the conversation so far since3744ChatGPT came out is about foundation models and how they3745operate and how they hallucinate, and some of the shortcomings3746of them and the risks with them, but a lot of these problems3747can be overcome at the application layer. And Ms. Miller and I3748represent companies who build applications on top of many3749models, and so if you think about that framework, what you3750understand then is the responsibility of some of these3751outcomes, the responsibility of some of these behaviors is not3752just left to the model. It is left to the vendor and the3753developer of----3754    Mrs. Luna. So, how you train it.3755    Mr. Shah. How you train it, how you package it, how you3756check for issues and guardrails and all of that.3757    Mrs. Luna. Yes. So, real quick, I have one more question,3758just real quick, yes or no. Would either or any of you be in3759favor of the U.S. Government through either the DIA, NSA, all3760agencies involved, Pentagon, even creating a super AI--I would3761maybe term it guardian AI--to actually train up, based on our3762principles, to defend the United States and potentially world3763against maybe a more nefarious actor like China, creating the3764same super weapon, because AI can only fight AI, I think you3765guys would all agree with that. In an effort to defend the home3766front?3767    Mr. Schneier. I think we cannot do that yet.3768    Mrs. Luna. Not yet, but maybe in the future?3769    Mr. Schneier. Maybe in the future.3770    Mrs. Luna. OK.3771    Mr. Thierer. Maybe, but I need more about the details about3772what that means because there is a lot to unpack there.3773    Ms. Miller. Unqualified, probably, to weigh in on that one.3774    Mr. Shah. I am supportive of any sort of sovereign AI that3775we think makes sense.3776    Mrs. Luna. Sovereign AI would be a good name.3777    Mr. Bajraktari. I think government alone cannot do it. I3778think public-private partnership should be required in this3779case because, otherwise, we will not be able to confront China.3780    Mrs. Luna. Thank you.3781    Ms. Greene. The gentlelady's time has expired. I now3782recognize Mr. Frost from Florida.3783    Mr. Frost. Thank you, Chair. You know, the first few months3784of the Trump Administration have been unhinged. At several3785events in my district, I have heard from outraged constituents.3786People, as I have traveled across the country, have heard the3787same from people in many different districts. And there are two3788common themes I keep hearing about: No. 1, people that are3789angry at the Administration's attempts to degrade vital3790services, like, Social Security, Medicaid, FEMA, and the VA;3791and also fear about what Elon Musk and folks at DOGE have done3792and could be doing with the data of American citizens and the3793American people, things like health records, banking3794information, and Social Security numbers. These conversations3795are really just the tip of the iceberg. Hundreds of3796constituents have also called into my office with the same3797fear, that their private information is falling into the wrong3798hands.3799    Mr. Schneier, reporting in May revealed that DOGE is now3800feeding unknown amounts and unknown types of sensitive3801government data into Grok, the AI program owned by Elon Musk. I3802find this extremely concerning. What are the potential harms3803that individuals now face from DOGE doing something like this?3804    Mr. Schneier. Well, we do not know. So, it is the two3805things I would worry about, feeding it to Grok and also giving3806it to Palantir to train their models. So, once it has now left3807the government, it is going to different companies for3808different things. And I think that the fact that we do not know3809what the data is, what is being done to it, how it is being3810used, how it is being protected, is a grave danger, right? We3811secure that data because it is important and now it is3812somewhere else also. Also, we do not know the accuracy. I have3813not talked about it. I would worry about data poisoning. If you3814were an American adversary, one of the things you might want to3815do is go to that ill-protected data and change stuff because3816now the model is being trained on false data and we talked a3817little bit about hallucinations and making bad decisions.3818    Mr. Frost. In what way could the dumping of Americans'3819private data into personal and corporate AI programs create3820conflicts of interest or create personal enrichment for certain3821people?3822    Mr. Schneier. Now, I mean, data is power and whoever has3823the data has the power, and for us to give that data to3824corporations, really with, you know, not even charging them for3825it, gives them an enormous amount of power. How they can use3826that is going to depend on what they are doing, and future3827technologies are going to make different things possible, but3828think of it broadly as it increases the power of those who have3829the data.3830    Mr. Frost. And it is important for everyone to, you know,3831remember that we have seen many different conflicts of interest3832throughout this Administration. We have also seen Trump, Musk,3833and other folks dismantle data privacy protections and ignore3834law after law. Since we cannot rely on the President or his3835Administration to obey the law, what other safeguards currently3836exist to keep our data safe?3837    Mr. Schneier. No, that is all you got, I mean, you as a3838citizen. I mean, my data, that is at the IRS. I can do nothing3839about it, right? I am relying on you, the government, to3840protect my data. If you are not doing that writ large, I am3841kind of screwed.3842    Mr. Frost. For 130 days, we have seen Musk and other actors3843lurk inside our most sensitive data bases, grabbing whatever3844private data they want with zero to little oversight,3845especially no oversight from the U.S. Congress, including3846Republican Members of Congress as well. How can we as the3847Oversight Committee assess the damage that has been done and3848hold people accountable?3849    Mr. Schneier. I think you actually have to do the work.3850There is an enormous amount of investigation that has to3851happen. We do not know what data was taken. We do not know3852where it went. We do not know where it is. And that is going to3853be work figuring it out, but I think it is work that has to be3854done because I think the first step to fixing the problem is3855understanding what it is. I mean, we can stop the flow, you can3856close down the loopholes, you can get these people out of the3857organizations, you can start protecting the data, but now the3858data is gone. You have to figure out where it is, who has it,3859and then take it back.3860    Mr. Frost. Interestingly enough, you know, you brought up3861the word ``accountability,'' which is really important. When3862the Administration changed over from Democrats to Republicans,3863I think little things matter. I think words matter, and, you3864know, there is a word that changed in the name of this3865Committee. The word ``accountability'' used to be in the name3866of this Committee, and they changed it from government3867accountability to ``government reform.'' It seems like a lot of3868my Republican colleagues have no interest in holding this3869Administration accountable for the misuse of data, and Congress3870has a horrible record in being able to legislate as it relates3871to technology. I think there are huge educational barriers. We3872bury our heads in the sand until the problem gets out of hand,3873and then someone comes up with a grand solution of banning it.3874This is what happened with social media, unfortunately, what I3875think might also happen with AI unless we get our stuff3876together here. Thank you. I yield back.3877    Ms. Greene. The gentleman yields. I now recognize Mr. Crane3878from Arizona.3879    Mr. Crane. Thank you for holding this hearing today. I want3880to say thank you to all the panelists for coming. I think we3881are all aware that the age of AI is here and expansion is3882absolute and unstoppable, but I do want to talk about what I3883feel to be the elephant in the room when we talk about AI, and3884one of my major concerns as a representative of the people to3885this country is our workforce.3886    Recently, Dario Amodei, CEO of Anthropic, one of the3887world's most powerful creators of AI, has a scary warning for3888all of us. He says this: ``AI could wipe out half of the entry-3889level white-collar jobs and spike unemployment 10 to 20 percent3890in the next 1 to 5 years.'' Other experts have said that number3891is closer to 50 percent in the next 5 years. He also said that3892``AI companies and the government need to stop sugarcoating3893what is coming, the possible elimination of jobs across3894technology, finance, law, consulting, and other white-collar3895professions, especially entry-level gigs,'' he says.3896    Steve Bannon, who was a top advisor to President Trump in3897his first term, recently told Axios, ``I don't think anyone is3898taking into consideration how much administration,3899administrative, managerial, and tech jobs for people under 30,3900entry-level jobs that are so important in your 20s, are going3901to be eviscerated, and it is already happening.'' Research that3902we have done shows that IBM just laid off 8,000 employees as of3903last week; McKinsey, a consulting firm, 5,000 employees as of3904May 28. Dell has laid off 6,000 employees in 2024. Intel has3905laid off 15,000 employees in 2024. And all of these were3906credited to AI and efficiency. Estimated total jobs lost,3907214,000 total jobs lost to AI.3908    I recently asked AI, what are the careers that will be3909replaced first by AI, and AI told me data entry clerks,3910telemarketers, retail cashiers, warehouse workers, paralegals3911and legal assistants, basic customer support representatives,3912bank tellers, fast food workers, IT support technicians, tax3913preparers. So, we are talking about tens of millions of jobs,3914and I think that that is very scary, even though we have been3915talking about the efficiencies of AI, how much money it can3916save the Federal Government, corporations, and how we have to3917keep up with China.3918    One of my questions to you guys, and I am not blaming any3919of you, but does it bother you at all that the technology you3920are developing is set to destroy tens of millions of jobs? I3921want to start with you, Mr. Shah.3922    Mr. Shah. You know, technology over time has generally3923raised the quality of living for all humans, and, you know, I3924know a lot of my fellow CEO AI folks out there try to grab3925headlines with some of their claims. I have not seen Dario's3926math behind all of this, but I think the point there is3927correct, that low-paying entry jobs will be taken on by3928computers, but it gives us a chance to reposition those3929individuals, re-skill them, and I do not think this is going to3930end. I think it is going to continue. For the next, you know,3931every 7 years, we are going to be talking about this.3932    What we have found is that for customers of ours, if you3933look at the private sector, they have taken those efficiencies,3934turned them into profits. Profits have then led to more3935ambition to grow their businesses. In fact, the majority of our3936customers have actually increased their employee headcount over3937time because they want to endeavor to do more things. So, I3938think when we start to automate mundane tasks, we can start to3939do more interesting work, and I think organizations will take3940full advantage of that.3941    Mr. Crane. And you think that applies across the board to3942all those careers and industries that I just listed off, not3943just the ones in tech?3944    Mr. Shah. I think that there will be changes to many3945industries because of this, and I think we are going to have3946to, as a species, respond to what this brings us and be able3947to, if we want to continue to lead the global race in AI, our3948agility is going to really matter.3949    Ms. Miller. I would just add----3950    Mr. Crane. Go ahead.3951    Ms. Miller. I would like to just add that, I mean, Mr.3952Schneier mentioned this earlier, this is a serious issue. We3953are going to displace a lot of people, and we do not have the3954training right now to give them new jobs. We are talking about3955unemployment rates of 15 percent of college graduates, to say3956nothing of those who have not got a college degree. So, I do3957believe this is a serious issue, and it is not going to get any3958better.3959    Mr. Crane. I have one more question. I want to shift gears3960real quick and talk about AI and defense. Mr. Bajraktari, I am3961sorry, I looked at your resume. You were the Executive Director3962of National Security Commission; Chief of Staff to National3963Security Advisor, Lieutenant General H.R. McMaster; former3964Chief of Staff to Deputy Secretary of Defense. Recently, the3965Ukrainians launched a drone attack on Russian soil. Ukraine3966said over 40 bombers, or about a third of Russia's strategic3967nuclear bomber fleet, was damaged. It was reported that this3968was done with AI. Do you think that this attack could have been3969possible without the U.S. help with AI intelligence?3970    Ms. Greene. The gentleman's time has expired, but go ahead3971and answer quickly.3972    Mr. Bajraktari. So, I do not have any insights into that. I3973left the government a couple of years ago, but it only shows3974how technology is changing the battlefield today, and I think3975how Ukrainians are adopting to this new battlefield is really3976an example that they have showcased last weekend.3977    Mr. Crane. Thank you. I yield back.3978    Ms. Greene. I now recognize Mr. Bell from Missouri.3979    Mr. Bell. Thank you, and thank you to our witnesses for3980being here today to speak on this important issue.3981    Americans are concerned about AI and want protections from3982the risks it poses to their privacy, security, rights, and3983freedoms. Key AI guardrails that provided these protections at3984the Federal level were repealed. Now House Republicans are3985pushing legislation that bans states from enforcing laws or3986regulations to protect Americans from AI for a full decade. Mr.3987Schneier, why is it important that the Federal Government adopt3988AI responsibly and with necessary oversight?3989    Mr. Schneier. Because doing it badly is going to harm3990people, and these are powerful things. So, these are3991technologies that are going to, in a sense, replace human3992activities. Now, whether they are going to be assistance to3993humans or they are going to replace humans for low-level3994decisions, they are going to do the things that humans did. And3995we know that when humans make bad decisions in government,3996people get hurt, so AI is going to do the same thing, and the3997problem is they can do it faster. They can do it more3998efficiently, so they could more efficiently hurt people, they3999could more efficiently help people, and that is why you need to4000do it responsibly.4001    Mr. Bell. When DOGE pressured Secretary Hegseth and the4002Department of Defense to quickly purge any mention of race4003across its web pages, the Agency used AI. The result was the4004Pentagon removed a tribute to Jackie Robinson, an American4005sports legend, civil rights icon, and World War II Army4006veteran, who served his country honorably, simply because it4007acknowledged his race. We are talking about American history4008that was flagged and removed in the blink of an eye. Mr.4009Schneier, without such safeguards, what risks do we face in4010ensuring the prevention of purges, distortions, or4011discrimination in Federal AI?4012    Mr. Schneier. So, again, it is less the AI and more the4013people that used it, and the weirder one is we removed the page4014about the Enola Gay bomber, which was an aircraft and not4015actually gay. It just had the name. So, when these technologies4016are used without human oversight, as humans, we could recognize4017that is ridiculous, but an AI does not have that context. An AI4018is going to do, like any computer, exactly what it is told, and4019if it is given bad instructions, if it is not monitored, if4020there is not audit, there is not integrity in the system, you4021are going to get bad outcomes. So, the AI did it, but blame the4022humans who asked the AI to do it.4023    Mr. Bell. Removing access to information is just the first4024step toward deeper consequences. DOGE is creating lasting4025impact in our expert Federal workforce alone. We have seen a4026purging of nonpartisan Federal workforce, even though the4027Government Accountability Office has for years been sounding4028the alarm that a lack of skilled workers will lead to more4029waste, fraud, and abuse in government. Understanding how AI4030works and its possible biases in programming requires a very4031specific skill set, one which we are losing every day. Mr.4032Schneier, again, why is it important to retain skilled workers4033with institutional knowledge who can carefully oversee the4034integration and use of AI across our Federal agencies?4035    Mr. Schneier. Because the AI is not ready to do that yet.4036Without the people, the AI is going to run without context, and4037you are going to get mistakes like the ones we just talked4038about. And these are mistakes that are, in some ways,4039theoretical, they are about information, but they also harm4040people. We are talking about using AI to reduce fraud. Now you4041are going to have problems with false negatives. It is fraud4042you are not going to catch. You are going to have problems with4043false positives. There is fraud you are going to catch that is4044not real. A really good example comes from the U.K. Horizon4045Software was used to audit rural post offices and flagged4046hundreds of instances of fraud that was not real, costing4047people their jobs. There were suicides. This is the problem of4048getting it wrong, and humans can make these same mistakes. We4049just make them slower. AI makes the pace of this much worse.4050    Mr. Bell. In the name of efficiency, the government has4051recently slashed funding for Job Corps, a program that offers4052free career technical training for young people aged 16 to 24.4053Instead of investing in our future workforce, they have4054eliminated a program that could help fill the shortage of4055skilled workers in emerging tech careers. When it comes to the4056Federal Government, it is not a question of just efficiency,4057but also of equity, fairness, and safety. I believe in4058efficiency, and I believe that technology can do enormous good4059for the American people, but this is a country that has a4060history that AI cannot erase. I look forward to ensuring that4061when the Federal Government employs AI, it actually serves the4062needs of those who have too often been left behind.4063    Ms. Greene. The gentleman's time has expired.4064    Mr. Bell. I yield back. Thank you.4065    Ms. Greene. I now recognize Mr. McGuire from Virginia.4066    Mr. McGuire. Thank you, Madam Chair. Thank you to our4067witnesses for being here today. I want to start by saying that4068I agree with Madam Chair, Congresswoman Marjorie Taylor Greene,4069that somehow we missed this 10-year ban on AI state laws, and I4070am hoping that we will fix that in the U.S. Senate. The4071President Trump and the White House, they said that we must4072ensure American AI dominance, and just yes or no real quick--4073just yes or no real quick--do you agree with the Trump4074Administration that the U.S. must have AI dominance? Just yes4075or no real quick.4076    Mr. Bajraktari. Yes.4077    Mr. Shah. Yes.4078    Ms. Miller. Yes.4079    Mr. Thierer. Yes.4080    Mr. Schneier. No, I do not think it works that way anymore.4081    Mr. McGuire. Thank you. All right. So, DOGE has done4082outstanding work in identifying waste, fraud, and abuse. They4083have brought to the public's attention extreme inefficiencies4084in our government, especially involving outdated technology. In4085one example, DOGE highlighted the OPM Retirement Operations4086Center. This facility is located in an underground mine that4087holds over 400 million individual paper documents. It should4088come as no surprise that using outdated technologies leads to4089more mistakes. During the Fiscal Year 2024, the Federal4090Government reported an estimated $162 billion in improper4091payments. Banks and credit card companies use AI and machine4092learning to prevent fraud, oftentimes before any money changes4093hands.4094    So, I guess I will ask each witness, do you think the4095Treasury payment system could use similar technology to lower4096waste, fraud, and abuse? And I am talking AI.4097    Mr. Bajraktari. Yes.4098    Mr. Shah. In general, yes. I am not qualified on this4099particular topic.4100    Ms. Miller. Well, there are a number of executive orders4101and legislative proposals that are aimed at trying to improve4102Treasury's payment verification services and the data that4103Treasury is being able to gather in. But one of the problems4104with focusing entirely on Treasury, though I do believe4105Treasury is the right entity for a centralized system, is that4106Treasury is making a decision right before a payment is made4107where we can actually put these technologies upstream within4108the agencies before an application is approved because many4109times we are approving people, we are approving the shell4110companies or entities that do not exist. So, if we are using4111better data across the agencies, not just at the Treasury, but4112absolutely modernizing Treasury system and getting more data,4113and I am very supportive of the executive orders that have come4114out to date and some of the legislative bills that are being4115proposed.4116    Mr. McGuire. I got a couple more questions coming. Let me4117ask the last two witnesses, yes or no.4118    Mr. Thierer. Yes.4119    Mr. Schneier. I am with Ms. Miller that you want to push4120this upstream.4121    Mr. McGuire. OK. Do you think the Federal Government is4122losing money by not using modern technology, including AI and4123machine learning? That is just a yes or no.4124    Mr. Bajraktari. It is more complicated. So, you need the4125digital infrastructure to use AI, so I think we need to first4126modernize the government IT infrastructure so you will be able4127to use AI.4128    Mr. Shah. The private sector is doing this, so I think the4129government could do as well.4130    Ms. Miller. Absolutely, without question.4131    Mr. Thierer. Yes, I agree.4132    Mr. Schneier. Yes, especially tax revenue.4133    Mr. McGuire. OK. Mr. Shah, as AI becomes more integrated in4134decision-making systems, are you concerned that these tools4135might reflect political bias? Specifically, are today's AI4136models biased toward left-leaning perspectives, and what4137implications might that have for public trust or government4138use?4139    Mr. Shah. There is a proliferation of AI models now being4140trained on all sorts of data sets, and hallucinations are a4141thing, bias are a thing, and we can never really overcome them4142forever. I think there are going to still be traces of that4143given the type of technology that they are, but you can build4144applications that can control for this. You can absolutely look4145and have models checking models, checking models, and have a4146structure that ensures that everything it decides can be backed4147up and that can be referenced. And while that will not solve4148every particular case, nor can we do that with human beings, I4149think the models and the AI systems can get very, very4150accurate.4151    Mr. McGuire. President Biden let in, well, 15 million4152illegal immigrants into our country and allowed deadly drugs,4153like fentanyl, to pour into our communities. In just under 64154months, President Trump has secured it with daily encounters4155dropping by 95 percent. We have seen Department of Homeland4156Security successfully deploy AI to stop illegal immigration4157from cameras and sensors along the border to drug detection4158equipment and ports of entry. However, severe damage has4159already been done. There are an untold number of illegal aliens4160currently reside in our country--and I apologize if I pronounce4161your name wrong--Mr. Thierer, do you see any ways we can4162leverage AI to help facilitate deportations?4163    Mr. Thierer. I have not followed that one closely,4164Congressman, but I do know DHS is using AI tools for these4165purposes.4166    Mr. McGuire. Awesome. Well, thank you for all your4167testimony today, and I yield back.4168    Ms. Greene. The gentleman yields. I now recognize Ms.4169Pressley from Massachusetts.4170    Ms. Pressley. Thank you. Just 2 weeks ago, Republicans4171passed a reconciliation bill that would ban all states from4172regulating AI for 10 years. Since most did not actually read4173the bill, they were unaware of that, so this is a timely4174hearing as Republicans are learning the error of their ways in4175real time. They are offering the American people up as guinea4176pigs for corporations to conduct AI experiments with no4177oversight, no regulation, and no enforcement.4178    Mr. Chair, I ask unanimous consent to enter into the record4179a letter by the Massachusetts Joint Committee on Advanced4180Information Technology, the Internet and Cybersecurity from May418130, 2025, which details the dangers of the Republican4182moratorium. Unanimous consent?4183    Mr. McGuire. [Presiding]. Without objection.4184    Ms. Pressley. All right. Now, let us talk about why4185regulating AI matters. Artificial intelligence reflects the4186assumptions of those who build it and the priorities of those4187who use it, and too often, those priorities fail to include the4188safety, the rights, and economic opportunity of Black4189communities and other marginalized groups. In my district, the4190Massachusetts 7th, we are proud to be a national hub for4191technology and innovation. For example, Northeastern4192University's Center for Inclusive Computing is leading the4193charge to make sure that AI opportunities are accessible to4194all. At the same time, we cannot allow AI to be the latest4195chapter in America's history of exploiting marginalized groups,4196namely the Black community. The government must invest in an4197approach rooted in equity that protects the rights of everyone.4198    That is why, Mr. Chair, I ask unanimous consent to enter4199into the record this report released last year from Color of4200Change titled, ``Black Tech Agenda: Advancing Equity and4201Reimagining Technology.''4202    Mr. McGuire. Without objection.4203    Ms. Pressley. All right. Mr. Schneier, do you know which4204government employee or agency or board is responsible for4205overseeing AI deployment across the executive branch to prevent4206violations of civil liberties and civil rights?4207    Mr. Schneier. I do not believe there is one, so if this is4208a test, I just failed.4209    Ms. Pressley. Well, the truth is no one knows. The4210responsibility of AI civil rights enforcement is not in4211anyone's job description, and that has got to change. The4212United States Federal Government is deploying AI and it is4213impacting our daily lives. There are more than 2,000 AI use4214cases across agencies. Right now, there is nothing stopping the4215Department of Health and Human Services from using AI systems4216to help decide who receives Medicaid, who receives food4217assistance, and disability benefits. If an algorithm flags4218someone incorrectly, a mother could quite literally lose access4219to her child's medication, or a family could lose food support.4220    There is nothing stopping the Department of Justice from4221using racially biased algorithms to decide who gets released4222from prison. AI should not be determining who gets care, who4223stays in prison, or who can feed their kids without any sense4224of oversight or recourse. The government's use of AI cannot be4225unchecked and unregulated. If we want the public to trust the4226Federal Government, then we must have transparency. I yield4227back.4228    Mr. McGuire. The gentlewoman yields. I now recognize4229Congressman Moskowitz for 5 minutes.4230    Mr. Moskowitz. Thank you. Thank you, Mr. Chair. I4231appreciate the waive-on today. You know, obviously, this is an4232important topic on artificial intelligence. I got to be honest.4233I definitely want to meet the person in the Capitol who was4234like, we are going to have a hearing on intelligence. Let us4235give that to Chairman Comer. But, you know, we saw some4236language wind up in a bill that everyone is complaining that,4237you know, they did not read. Remember, they were the ones who4238had the 72-hour rule. That was their rule. In fact, they cried4239and demanded that every bill come forward and be given 72 hours4240to read a bill. They complained when Democrats were in control.4241Meanwhile, they passed this bill in the dead of night. Do not4242listen to me. That is what Elon Musk is now saying on Twitter.4243    Yesterday, in the DOGE Subcommittee hearing, Chairman Comer4244announced new investigations into the Biden Administration, and4245I thought, of course. This makes total sense, right? We have4246recently brought back measles and the dire wolf and the tariff4247policies of the 1930s, so you know, why not resurrect the4248failed Biden impeachment? I do not want to go down memory lane4249about what happened last year, but, you know, they said they4250would call for impeachment. That is what they said, and they4251never did. After hundreds of interviews, fundraising emails,4252and speeches, and lying to their base, they never called for4253impeachment. So, I want to do the do not listen to what they4254say, watch what they do, and this is, of course, all a4255distraction, right?4256    They said they would lower food costs. They said they would4257take us into the golden age. They said they would end the war4258in Ukraine and Gaza. They said they would not touch Medicaid.4259They said they would lower interest rates. They said they would4260lower our debt. They said they would release the Epstein files.4261Remember that? This is not one of the ones that Pam gave out. I4262made my own. It is not hard, right? And the biggest one, the4263biggest one, my favorite, they said they would make government4264more efficient, DOGE. ``E'' stood for efficiency.4265    Name one department, one government service, anything that4266the government does that they made more efficient. Certainly4267not the Newark airport. Definitely not FEMA. We got an4268Administrator who is finding out about hurricane seasons, a4269little late. Total failure. No new technology, no new4270procurement rules. Procurement is a big problem in government.4271You got different agencies, different departments making4272different procurements, buying technology that cannot4273communicate with each other. By the way, we learned that after42749/11 in law enforcement and we still have not fixed that4275problem. No new streamlined processes. Zero, OK?4276    And so, look, sometimes people are visual learners, right?4277They are visual learners, and so look, this is what the4278Republicans promised the American people. They promised them4279this, OK? This was the Willy Wonka thing in London. This is4280what they promised them. This is what it was supposed to look4281like, right? The golden age, and that is not what they have4282delivered. What they have delivered is this. This is what they4283have delivered the American people. This is what is left of the4284DOGE effort that--she got to keep her job. This is what is4285left. And so, they are very good at telling the American people4286what they want to do, what they say they are going to do, but4287the American people should start watching what they are4288actually doing, or what they have done or what they have not4289done.4290    And so, listen, fine, bring in the Biden people again. We4291will go through this whole thing. Tens of millions of dollars4292wasted in the last Congress. We will do it all again, right,4293back to our old tricks, but of course, they want to bury them4294in the basement again. It is being proposed that, oh, we will4295take their depositions, bury them in the basement. Then they4296will leak the transcripts. They tell you, well, we want to4297shine transparency on the end of the Biden Administration. We4298want to get to the truth, but they want to hold these meetings4299in secret, not in front of the cameras, not in front of the4300American people, in the bottom of a basement. By the way, I4301noticed that the Oversight Hearings are now in the bottom of a4302basement. Ironic, after all the basement talk last Congress,4303but they just want to suppress information. They want to do4304selective leaks and they want to distract that this is the4305least productive Congress in 70 years.4306    So, listen, it is fine. We will do it all again, OK? Will4307not help people at the grocery store, will not help people save4308their healthcare. It will do none of that, but it will get you4309on OANN and Fox News and raise money. Thank you.4310    Mr. McGuire. I now recognize Congresswoman Trahan for 54311minutes.4312    Ms. Trahan. Well, thank you, Mr. Chairman. I appreciate you4313allowing me to be part of this important conversation. Over on4314the Energy and Commerce Committee, which is where I usually4315serve, we have a lot of conversations about technology, and one4316thing is always clear: data is at the heart of AI. That is why4317I believe that any serious discussion about AI has to start4318with a conversation about privacy, and that is what I am here4319to do today, to sound the alarm about a deeply troubling trend,4320our own government's growing appetite for Americans' personal4321data.4322    Let me just give you an example, a hypothetical, of course,4323but not a far-fetched one. Sarah is a typical American woman.4324She pays her taxes, owns a gun legally, and is raising her4325daughter, Emma, on her own. She and Emma rely on Medicaid to4326get the healthcare that they need. One day, Sarah shares a post4327on Facebook. She is concerned about something the President4328said about firearms and she posts so, but in Washington, an AI-4329powered monitoring system flags her post. A political appointee4330digs into her personal data and sends emails to agency heads4331urging them to take action against her. Within days, Sarah's4332life falls apart. The IRS audits her and claims she owes4333thousands. Emma's doctor says her Medicaid is not active4334anymore and now Sarah has to pay out of pocket. Now, to be4335clear, this story is made up, but it is not science fiction. It4336is an alarm. It is a warning.4337    Mr. Schneier, you talked in your testimony about coercion4338as an adversarial use of data. What kinds of coercion could bad4339actors inside the government use if they had detailed profiles4340on every American?4341    Mr. Schneier. I would think of it as selective4342investigation. I mean, the government has enormous powers to4343investigate people, and the question is who they choose to4344investigate. There is a famous book from many years ago, called4345``Three Felonies a Day,'' that we, in our normal lives, commit4346three felonies a day because there are just so many rules and4347we do not know them, and so, given things like that, who you4348choose to enforce the law on matters. So, this data can be used4349to select people whom to investigate, people whom to charge,4350and this could be used selectively by any regime, even not the4351U.S., any country that wants to do this.4352    Ms. Trahan. Yes. Well, unfortunately, this is not just a4353hypothetical trend. It is already happening. Under the Trump4354Administration, DOGE aggressively collected sensitive data4355across agencies, breaking down firewalls that are supposed to4356protect us. Then came the executive order directing agencies to4357eliminate information silos, basically to share and pool that4358data. And just last week, we learned that Palantir, a Silicon4359Valley company known for building surveillance tools, is being4360hired to build AI-powered profiles on every American using the4361data that DOGE collected. It is hard to overstate how dangerous4362this is. Mr. Schneier, are you worried that once this data is4363centralized, future administrations, no matter their party,4364could weaponize it? I mean, are we on the verge of opening4365Pandora's box here?4366    Mr. Schneier. I do not know if Pandora's box has been4367opened years ago, but certainly giving this power to a4368government is something that feels very un-American. I mean,4369there are reasons why this data was siloed. There are reasons4370why we did not have these powers. I mean, you can imagine4371humans doing this well before AI, but we chose not to. So, AI4372can certainly make this more efficient, but yes, this is power4373in the hands of a human who wants to wield it for ill, can do4374that very efficiently.4375    Ms. Trahan. Thank you. We need a national reckoning on4376privacy. That means strong oversight of this Administration and4377its tech partners and real legislation to protect Americans'4378rights. You know, I have spent the past 3 months talking with4379civil liberties groups, privacy experts, and people across the4380country, and the one thing is clear. We need stronger privacy4381laws. I believe that we can protect people's data and modernize4382government to prevent fraud, waste, and abuse. These goals are4383not at odds. They are linked. So, if you are listening and you4384are concerned about what is happening about Big Tech, about4385government overreach, about your family's privacy, call my4386office. Let us have a national conversation. Let us protect the4387freedom that our founders fought for and the privacy we all4388deserve.4389    And one last thing I just wanted to mention, because over4390the course of this hearing, the Chair has suggested that no one4391on the other side of the aisle called attention to the harms of4392the Republicans' 10-year ban on state AI regulations. That is4393patently false. We had robust debate on the Energy and Commerce4394Committee with several Democratic members, myself included,4395calling attention to this provision during and after our 26-4396hour markup. In fact, Democrats offered an amendment to strike4397the language entirely. So, Mr. Chair, I ask unanimous consent4398to enter into the record the results of the recorded vote.4399    Mr. McGuire. The gentlewoman's time is up. Without4400objection.4401    Ms. Trahan. Thank you. I yield back.4402    Mr. McGuire. In closing, I want to thank the panelists once4403again for their testimony today.4404    With that and without objection, all Members will have 54405legislative days within which to submit materials and to submit4406additional written questions for the witnesses, which will be4407forwarded to the witnesses for their response.4408    If there is no further business, without objection, the4409Committee stands adjourned.4410    [Whereupon, at 2 p.m., the Committee was adjourned.]44114412                                 [all]