[House Hearing, 119 Congress]
[From the U.S. Government Publishing Office]
ARTIFICIAL INTELLIGENCE: EXAMINING TRENDS
IN INNOVATION AND COMPETITION
=======================================================================
HEARING
BEFORE THE
SUBCOMMITTEE ON THE ADMINISTRATIVE STATE,
REGULATORY REFORM, AND ANTITRUST
COMMITTEE ON THE JUDICIARY
U.S. HOUSE OF REPRESENTATIVES
ONE HUNDRED NINETEENTH CONGRESS
FIRST SESSION
__________
WEDNESDAY, APRIL 2, 2025
__________
Serial No. 119-13
__________
Printed for the use of the Committee on the Judiciary
[GRAPHIC NOT AVAILABLE IN TIFF FORMAT]
Available via: http://judiciary.house.gov
__________
U.S. GOVERNMENT PUBLISHING OFFICE
59-906 WASHINGTON : 2025
-----------------------------------------------------------------------------------
COMMITTEE ON THE JUDICIARY
JIM JORDAN, Ohio, Chair
DARRELL ISSA, California JAMIE RASKIN, Maryland, Ranking
ANDY BIGGS, Arizona Member
TOM McCLINTOCK, California JERROLD NADLER, New York
THOMAS P. TIFFANY, Wisconsin ZOE LOFGREN, California
THOMAS MASSIE, Kentucky STEVE COHEN, Tennessee
CHIP ROY, Texas HENRY C. ``HANK'' JOHNSON, Jr.,
SCOTT FITZGERALD, Wisconsin Georgia
BEN CLINE, Virginia ERIC SWALWELL, California
LANCE GOODEN, Texas TED LIEU, California
JEFFERSON VAN DREW, New Jersey PRAMILA JAYAPAL, Washington
TROY E. NEHLS, Texas J. LUIS CORREA, California
BARRY MOORE, Alabama MARY GAY SCANLON, Pennsylvania
KEVIN KILEY, California JOE NEGUSE, Colorado
HARRIET M. HAGEMAN, Wyoming LUCY McBATH, Georgia
LAUREL M. LEE, Florida DEBORAH K. ROSS, North Carolina
WESLEY HUNT, Texas BECCA BALINT, Vermont
RUSSELL FRY, South Carolina JESUS G. ``CHUY'' GARCIA, Illinois
GLENN GROTHMAN, Wisconsin SYDNEY KAMLAGER-DOVE, California
BRAD KNOTT, North Carolina JARED MOSKOWITZ, Florida
MARK HARRIS, North Carolina DANIEL S. GOLDMAN, New York
ROBERT F. ONDER, Jr., Missouri JASMINE CROCKETT, Texas
DEREK SCHMIDT, Kansas
BRANDON GILL, Texas
MICHAEL BAUMGARTNER, Washington
------
SUBCOMMITTEE ON THE ADMINISTRATIVE STATE,
REGULATORY REFORM, AND ANTITRUST
SCOTT FITZGERALD, Wisconsin, Chair
DARRELL ISSA, California JERROLD NADLER, New York, Ranking
BEN CLINE, Virginia Member
LANCE GOODEN, Texas J. LUIS CORREA, California
HARRIET HAGEMAN, Wyoming BECCA BALINT, Vermont
MARK HARRIS, North Carolina JESUS G. ``CHUY'' GARCIA, Illinois
DEREK SCHMIDT, Kansas ZOE LOFGREN, California
MICHAEL BAUMGARTNER, Washington HENRY C. ``HANK'' JOHNSON, Jr.,
Georgia
CHRISTOPHER HIXON, Majority Staff Director
JULIE TAGEN, Minority Staff Director
C O N T E N T S
----------
Wednesday, April 2, 2025
OPENING STATEMENTS
Page
The Honorable Scott Fitzgerald, Chair of the Subcommittee on the
Administrative State, Regulatory Reform, and Antitrust from the
State of Wisconsin............................................. 1
The Honorable Jerrold Nadler, Ranking Member of the Subcommittee
on the Administrative State, Regulatory Reform, and Antitrust
from the State of New York..................................... 3
The Honorable Jim Jordan, Chair of the Committee on the Judiciary
from the State of Ohio......................................... 4
WITNESSES
Neil Chilson, Head, AI Policy, Abundance Institute
Oral Testimony................................................. 6
Prepared Testimony............................................. 8
Joseph V. Coniglio, Director, Antitrust and Innovation Policy;
Schumpeter Project, Competition Policy, Information Technology
and Innovation Foundation
Oral Testimony................................................. 20
Prepared Testimony............................................. 22
Jessica Melugin, Director, Center for Technology and Innovation,
Competitive Enterprise Institute
Oral Testimony................................................. 28
Prepared Testimony............................................. 30
Alvaro M. Bedoya, Former Commissioner, Federal Trade Commission
(FTC)
Oral Testimony................................................. 33
Prepared Testimony............................................. 35
LETTERS, STATEMENTS, ETC. SUBMITTED FOR THE HEARING
All materials submitted for the record by the Subcommittee on the
Administrative State, Regulatory Reform, and Antitrust are
listed below................................................... 63
Materials submitted by the Honorable J. Luis Correa, a Member of
the Subcommittee on the Administrative State, Regulatory
Reform, and Antitrust from the State of California, for the
record
An article entitled, ``China's top universities expand
enrolment to beef up capabilities in AI, strategic
areas,'' Mar. 10, 2025, The Economic Times
An article entitled, ``65% Of Top AI Companies Have Immigrant
Founders,'' Jul. 9, 2023, Forbes
An article entitled, ``Trump's Science Policies Pose Long-
Term Risk, Economists Warn,'' Mar. 31, 2025, The New York
Times
An article entitled, ``Trump Administration Has Begun a War
on Science, Researchers Say,'' Mar. 31, 2025, The New
York Times
An article entitled, ``China leads world in generative AI
adoption, underscoring country's progress,'' Jul. 10,
2024, Reuters
An article entitled, ``Former Google Employees Create 14 Of
The Top 50 AI Startups,'' Dec. 9, 2024, Performance
Marketing Insider
An article entitled, ``Trump Loves AI But His Tariffs Ramp Up
Costs for Data Centers,'' Mar. 19, 2025, Bloomberg
An article entitled, ``Apple chip engineer returns to China,
joins Fudan University amid push for talent,'' Mar. 31,
2025, South China Morning Post
An article entitled, ``Immigrant Entrepreneurs Bring Jobs And
Innovation, New Research Shows,'' May 23, 2024, Forbes
An article entitled, ``From underdog to genius: Yale computer
scientist Sun Huanbo returns to China,'' Mar. 30, 2025,
South China Morning Post
An article entitled, ``Tech companies are telling immigrant
employees on visas not to leave the U.S.,'' Mar. 31,
2025, Washington Post
Materials submitted by the Honorable Jerrold Nadler, Ranking
Member of the Subcommittee on the Administrative State,
Regulatory Reform, and Antitrust from the State of New York,
for the record
An article entitled, ``AI Watch: Global regulatory tracker--
China,'' Apr. 1, 2025, JD Supra
An article entitled, ``US-China AI Race: AI Policy as
Industrial Policy,'' Apr. 11, 2023, AI Now Institute
An article entitled, ``Elon Musk, who co-founded firm behind
ChatGPT, warns A.I. is `one of the biggest risks' to
civilization,'' Feb. 15, 2023, CNBC
An article entitled, ``OpenAI peels back ChatGPT's safeguards
around image creation,'' Mar. 28, 2025, Tech Crunch, submitted
by the Honorable Jesus G. ``Chuy'' Garcia, Member of the
Subcommittee on the Administrative State, Regulatory Reform,
and Antitrust from the State of Illinois, for the record
Materials submitted by the Honorable Jim Jordan, Chair of the
Committee on the Judiciary from the State of Ohio, for the
record
An article entitled, ``The FTC must be held accountable for
its widespread leaks of confidential data,'' Oct. 19,
2024, The Hill
A letter to the Honorable Jim Jordan, Chair of the Committee
on the Judiciary from the State of Ohio, Aug. 26, 2024,
from Mark Zuckerberg, Founder, Chairman & CEO, Meta
Platforms, Inc.
Materials submitted by the Honorable Scott Fitzgerald, Chair of
the Subcommittee on the Administrative State, Regulatory
Reform, and Antitrust from the State of Wisconsin, for the
record
A letter to the Honorable Scott Fitzgerald, Chair of the
Subcommittee on the Administrative State, Regulatory
Reform, and Antitrust from the State of Wisconsin, and
the Honorable Jerold Nadler, Ranking Member of the
Subcommittee on the Administrative State, Regulatory
Reform, and Antitrust from the State of New York, Apr. 8,
2025, from Morten C. Skroejer Senior Director, Technology
Competition Policy, SIIA Policy
A letter to the Honorable Jim Jordan, Chair of the Committee
on the Judiciary from the State of Ohio; the Honorable
Jamie Raskin, Ranking Member of the Committee on the
Judiciary from the State of Maryland; the Honorable Scott
Fitzgerald, Chair of the Subcommittee on the
Administrative State, Regulatory Reform, and Antitrust
from the State of Wisconsin; and the Honorable Jerrold
Nadler, Ranking Member of the Subcommittee on the
Administrative State, Regulatory Reform, and Antitrust
from the State of New York, Apr. 2, 2025, from Gary
Shapiro, CEO, Vice Chair, Consumer Technology
Association, and Kinsey Fabrizio, President, Consumer
Technology Association
A statement from Jennifer Huddleston, Technology Policy
Senior Fellow, CATO Institute, Apr. 2, 2025
A letter to the Honorable Scott Fitzgerald, Chair of the
Subcommittee on the Administrative State, Regulatory
Reform, and Antitrust from the State of Wisconsin, and
the Honorable Jerold Nadler, Ranking Member of the
Subcommittee on the Administrative State, Regulatory
Reform, and Antitrust from the State of New York, Apr. 1,
2025, from Graham Dufault, General Counsel, ACT | The App
Association
An article entitled, ``DOGE Plan to Push AI Across the US Federal
Government is Wildly Dangerous,'' Mar. 6, 2025, Tech Policy
Press, submitted by the Honorable Jesus G. ``Chuy'' Garcia, a
Member of the Subcommittee on the Administrative State,
Regulatory Reform, and Antitrust from the State of Illinois,
for the record
APPENDIX
Materials submitted by the Honorable J. Luis Correa, a Member of
the Subcommittee on the Administrative State, Regulatory
Reform, and Antitrust from the State of California, for the
record
An article entitled, ``Australia Urges Universities to
Diversify Research Away From US,'' Apr. 1, 2025,
Bloomberg
An article entitled, ``Trump's Trade War Pushes Canadian Tech
Workers to Rethink Silicon Valley,'' Mar. 31, 2025, Wired
QUESTIONS AND RESPONSES FOR THE RECORD
Questions submitted by the Honorable J. Luis Correa, a Member of
the Subcommittee on the Administrative State, Regulatory
Reform, and Antitrust from the State of California, for the
record
Questions to Alvaro Bedoya, Former Commissioner, Federal Trade
Commission, Apr. 2, 2025
Questions to Neil Chilson, Head, AI Policy, Abundance
Institute, Apr. 2, 2025
Response to questions from Neil Chilson, Head, AI Policy,
Abundance Institute, Oct. 24, 2024
Questions to Joseph V. Coniglio, Director, Antitrust and
Innovation Policy; Schumpeter Project, Competition Policy,
Information Technology and Innovation Foundation, Apr. 2,
2025
Response to questions from Joseph V. Coniglio, Director,
Antitrust and Innovation Policy; Schumpeter Project,
Competition Policy, Information Technology and Innovation
Foundation, May 1, 2025
Questions to Jessica Melugin, Director, Center for Technology
and Innovation, Competitive Enterprise Institute, Apr. 2,
2025
Response to questions from Jessica Melugin, Director, Center
for Technology and Innovation, Competitive Enterprise
Institute, Apr. 17, 2025
ARTIFICIAL INTELLIGENCE: EXAMINING TRENDS IN INNOVATION AND COMPETITION
----------
Wednesday, April 2, 2025
House of Representatives
Subcommittee on the Administrative State,
Regulatory Reform, and Antitrust
Committee on the Judiciary
Washington, DC
The Subcommittee met, pursuant to notice, at 10:02 a.m., in
Room 2141, Rayburn House Office Building, the Hon. Scott
Fitzgerald [Chair of the Subcommittee] presiding.
Present: Representatives Fitzgerald, Cline, Gooden,
Hageman, Harris, Nadler, Correa, Balint, Garcia, and Johnson.
Also present: Representative Kiley.
Mr. Fitzgerald. I want to thank all the Members for being
here. The Subcommittee will come to order. Without objection,
the Chair is authorized to declare a recess at any time. We
welcome everyone to today's hearing on artificial intelligence
and trends in innovation and competition. I will now recognize
myself for an opening statement.
Artificial intelligence, AI, is a powerful new technology,
and it's moving fast. It already is changing the way businesses
operate, how people work, and how we solve problems throughout
our society. From agriculture to manufacturing to customer
service, AI is showing up in more places every day, and it's
helping Americans get more done faster and more efficiently.
There's no doubt this is a big moment for both America and
the rest of the world, but the question before us today is
this: Will we respond to it with more freedom or with more
government control?
America has always led the way in innovation, not because
we have the biggest government, but because we have trust in
our people. We believe in free markets, competition, and
limited government. That's what built our economy, and that's
how we've created new technologies from the light bulb to the
smartphone, and that's why the United States--not China or
Europe--has been the global leader in innovation for over a
century. That same model is working in AI.
The story we hear sometimes is that a few big companies are
dominating the AI sector. When you look closer, the story
unravels throughout this industry. There is real competition
happening at every layer of the AI sector, in cloud
infrastructure, in AI models, and in tools and applications.
It's true that big tech plays a role in this sector, but to
also have startups, open-source developers, university labs,
and solo engineers pushing the boundaries of this exciting new
technology. Small businesses are using AI to do things they
never could afford before: Automate routine tasks, improve
customer service, analyze data, write content, design products,
and more.
AI is helping the little guy compete with the big guys.
It's lowering the barriers to entry across the board. This
isn't a monopoly story. It's a success story. It's the free
market doing what it does best.
Unfortunately, some in government want to step in and start
regulating before they even understand what they are
regulating. They want to create sweeping new rules, or even
whole new agencies to oversee AI under the claim of safety and
of fairness. They want to decide which models are, quote,
``responsible,'' which tools are, quote, ``approved,'' and
which approaches are fair and who gets to build what. That's
very dangerous.
We see this same saga playing out time and time again.
Overregulation can kill innovation. When the government inserts
itself too early, it locks in incumbents, drives up costs, and
freezes the market, startups get squeezed, the rules benefit
the insiders, and innovation slows to a crawl.
Meanwhile, our competitors, especially in China, aren't
waiting around. They're pouring billions into AI development.
They're using it to monopolize their military, monitor their
citizens, and try to leapfrog the U.S. on the global stage to
the detriment of Democrat principles and values.
Over in Europe, the EU is rolling out a top-down AI
regulatory regime that already is causing problems for American
companies trying to do business overseas. The rules are vague,
burdensome, and constantly shifting. They've been written by
bureaucrats who aren't accountable certainly to the American
people.
We should not copy China's model of control, and we should
not copy Europe's model of overregulation. We need to stay true
to what works, and that is free enterprise, open competition,
and light-touch regulatory approach that allows innovation to
flourish.
That doesn't mean we ignore real problems. We should
enforce our existing antitrust laws if companies are abusing
their market power. We should be on guard against fraud, abuse,
and national security threats. We don't need a heavy-handed new
regulatory regime to do that.
What we need is to keep the U.S. the best place in the
world, to build, test, and deploy AI. That means making room
for new entrants, it means protecting free speech in the
digital age, and it means making sure American entrepreneurs,
not unelected regulators, are the ones shaping the future of
this technology.
AI is a tool. Whether it helps or harms depends on how we
use it, and whether we keep the freedom to innovate, adapt, and
compete. The worst thing we can do is let fear drive us into
giving up control to Washington, to Brussels, or to Beijing.
The American model has always been the best model. Let's stick
to it.
I want to thank the witnesses for appearing before us
today, and I look forward to hearing what each of you has to
say.
I now recognize the Ranking Member, Mr. Nadler, for his
opening statement.
Mr. Nadler. Thank you, Mr. Chair. Thank you to our
witnesses, particularly to Commissioner Bedoya for joining us
today.
As we examine innovation and competition with respect to
artificial intelligence, we must keep in mind three simple
truths:
First, there is no artificial intelligence exception to the
antitrust and consumer protection laws.
Second, there is no DOGE exception to Federal privacy
protections.
Third, there is no Presidential exception to following
Federal laws.
It is a sad comment on the time we find ourselves in that
these undeniable and basic propositions must be continually
affirmed. AI technology is rapidly advancing and promises to
transform our society in ways we cannot even imagine today.
While AI will deliver many economic and social benefits, it
also poses tremendous risks that we are just beginning to
grapple with. With control of AI technology in the hands of
just a few entrenched companies, it is important that we keep a
watchful eye on the industry, with thoughtful regulation that
will foster growth and innovation while protecting consumers.
One of the most critical tools in this effort is antitrust
and consumer protection enforcement, led by the Federal Trade
Commission, an independent administrative agency that was
created more than 110 years ago by Congress as the Nation was
beset by powerful monopolies that dominated core industries and
society. Its specific mission is protecting the public from
deceptive or unfair business practices and from unfair methods
of competition.
Its structure is deliberately bipartisan with no more than
three members of the five-person commission able to be from a
single party. This makeup, along with the fact that the
commissioners are meant to serve staggered seven-year terms,
and are removable only for very specific causes, is designed to
increase the independence, and, thereby, the prestige of the
Commission.
We need the FTC now more than ever as we confront an AI-
induced wave of competition, consumer protection, and privacy-
related issues. Yet, the Trump Administration is undermining
the single agency best positioned to address them.
Years before ChatGPT burst onto the scene, the FTC was
already working to protect consumers by stopping the largest
semiconductor chip merger in history. Chips are an essential
input to AI, and blocking this merger helped protect innovation
by preserving competition for this vital tool. Using their
unique power as part of the Commission's study function, the
FTC also produced the report examining AI investments and
partnerships that would best enable the agency to ensure that,
when dominant companies pursued these practices, they did not
undermine competition and innovation.
With more and more companies relying on AI to make
fundamental decisions that affect consumers, the FTC has also
carefully examined the algorithms that power these systems. In
particular, the commissioners tackled algorithmic bias and
algorithmic collusion, which is when a computer system
systematically produces unfair outcomes or is used to engage in
anticompetitive pricing, respectively.
This work is essential to ensure that as AI develops, it
serves the American people. If we are serious about promoting
competition and innovation in AI, we need a strong FTC to
enforce the law. Instead, the Trump Administration has done all
it can to hobble the agency and to weaken enforcement of
antitrust and consumer protection laws.
Even more troubling, while with one hand, the
administration is undermining regulation of AI, with the other
hand it is handing massive troves of sensitive data on millions
of Americans to Elon Musk and his DOGE team that could be used
to deploy AI throughout the government. Governmental processes,
their workflows, and the use and interpretation of government
data are highly complex. There is scant evidence, however, that
AI can be successfully deployed in government on such a far-
reaching scale.
Musk and DOGE seem intent on forcing the American people to
embark on a grand experiment based on little more than their
faith in AI technology, an experiment that places all our
privacy and security at great risk.
While all this unfolds, Commissioner Bedoya should be
participating in any number of matters protecting consumers.
Even as a minority commissioner, he and his stellar Democratic
commissioner, Rebecca Slaughter, occupy an important and
congressionally mandated role. They are fundamental to infusing
commission deliberations and decisions with a bipartisan
dimension.
Oftentimes, minority commissioners influence the outcome or
logic of the decision. Even when they do not, they perform the
essential function of providing transparency into the process,
including by highlighting key concerns in their dissents.
Instead of working to enhance competition in the AI
industry, and ensuring that proper safeguards are in place,
they're forced to contest their blatantly unlawful termination
by President Trump. This unprecedented power grab followed
similar efforts to remove Democratic Members of other
independent agencies without cause.
As one court noted, these efforts, quote,
Threatened to upend fundamental protections in our
Constitution, but ours is not an autocracy. It is a system of
checks and balances.
Just as unchecked power by monopolies in the marketplace does
not secure the best outcomes for consumers, unchecked
government power does not secure the best outcomes for our
democracy. We should be mindful of both points as we begin
today's hearings.
I look forward to hearing from our witnesses, and I yield
back my time.
Mr. Fitzgerald. The gentleman yields back.
I now recognize Chair of the Full Committee, Mr. Jordan,
for his opening statement.
Chair Jordan. Thank you, Mr. Chair. American innovators
have always led the world, because we've always protected free
markets, limited government, and individual liberty. When it
comes to AI, the Biden Administration turned its back on
innovators and entrepreneurs.
Rather than trusting American ingenuity, the Biden
Administration tried to put Washington in charge, empowering
unelected and unaccountable bureaucrats. Who paid the price?
The very people we should be empowering. The startups trying to
challenge big tech, the researchers building open models, the
platforms that actually support free speech.
Meanwhile, around the world, we see the dangers of pursuing
the wrong approach to AI innovation. We see China pouring
billions of dollars into AI development to weaponize the
technology. We see Europe smothering innovation with
overregulation and impossible compliance burdens.
We should be clear; America must lead on AI. To do that, we
need to protect free speech, preserve real competition, and get
the government out of the way. That means (1) no back-door
censorship; (2) no weaponizing antitrust law to reward
political allies and punish dissent; and (3) letting no
international bodies or unelected regulators set the rules for
American technology.
Thankfully, President Trump--we have a President focused on
unleashing American innovation, on cutting red tape, and on
protecting free expression, and on making sure AI development
happens here, not in China, not in Europe. Under President
Trump's leadership, the focus is right where it should be:
Keeping America competitive, preserving free speech, and
getting the government out of the way.
Today's hearing is about how we protect innovation, protect
liberty, and ensure that AI remains a force for freedom, not
for control.
I want to thank our Chair for calling this hearing, and I
look forward to hearing from our witnesses. I yield back.
Mr. Fitzgerald. Gentleman yields back.
Without objection, all other opening statements will be
included in the record.
We will now introduce today's witnesses.
Mr. Neil Chilson. Mr. Chilson is the head of AI policy at
the Abundance Institute, a nonprofit organization focused on
emerging technology. He previously served as the Chief
Technologist at the Federal Trade Commission and as an attorney
adviser to then-FTC Acting Chair, Maureen Ohlhausen.
Mr. Joe Coniglio. Mr. Coniglio is the Director of Antitrust
and Innovation at the Information Technology and Innovation
Foundation where he leads the Schumpeter project on competition
policy. He focuses on digital platform, monopolization policy,
and dynamic competition.
Ms. Jessica Melugin. Ms. Melugin is the Director of the
Center for Technology and Innovation at the Competitive
Enterprise Institute. Her research focuses on technology
issues, including antitrust, online privacy, artificial
intelligence, telecommunications, and other issues.
Mr. Alvaro Bedoya. Mr. Bedoya is the former Commissioner at
the Federal Trade Commission. Prior to his appointment to the
FTC, he founded the Center on Privacy and Technology at the
Georgetown University Law School.
We welcome our witnesses and thank them for appearing
today. We will begin by swearing you in.
Would you please rise and raise your right hand.
Do you swear or affirm under penalty of perjury that the
testimony you're about to give is true and correct to the best
of your knowledge, information, and belief, so help you God.
Let the record reflect that the witnesses have answered in
the affirmative. Thank you and please be seated.
Please know that your written testimony will be entered
into the record in its entirety. Accordingly, we ask that you
summarize your testimony in five minutes.
Without objection, Mr. Kiley will be permitted to
participate in today's hearing for the purpose of questioning
the witnesses if a Member of the Subcommittee yields him time.
Please know that your written testimony will be entered in
the in its entirety. Mr. Chilson, you may begin.
STATEMENT OF NEIL CHILSON
Mr. Chilson. Thank you, Chair Fitzgerald, thank you to the
Chairs and Ranking Members of the Full Committee, and to
Ranking Member Correa, and the distinguished Members of the
Subcommittee.
Thank you for inviting me to testify today. I'm Neil
Chilson. I'm the head of AI Policy at the Abundance Institute.
We are a nonprofit dedicated to creating policy and cultural
environments that allow emerging technologies to thrive and
expand human prosperity.
Last year, American AI startups raised more than $97
billion or approximately half of the entire Apollo program. At
least 49 different U.S. AI startups raised more than $100
million just last year. These are record numbers, and they're
just getting started.
I'm here to make a simple but important point.
Economically, AI isn't an industry or a market. It's an
ecosystem of thousands of companies, and the U.S. AI ecosystem
is vibrant and remarkably competitive at every level. From
chips to cloud infrastructure, from foundation models to
applications, we're seeing intense competition driving rapid
innovation and democratizing access to powerful AI
capabilities.
Consider that 11 different developers last year globally--
across the globe achieved a GPT-4 level model, one of the most
advanced types of AI models; or that market shares in cloud
computing continued to shift with established players and
startups pressuring each other. The AI enterprise service
market is highly fragmented, the leading firm holding a mere
seven percent.
These are not signs of concentrated power. They're evidence
of healthy competition. This competitive environment is
spreading the benefits of AI widely. AI capabilities that were
cutting-edge, or even unimagined just a few years ago, are now
widely accessible or even free.
For example, many of you may have seen the most recent
generation of Studio Ghibli themes online. This type of use of
powerful image generation which once required specialized
knowledge and powerful computers is now available for anyone
for free or maybe pennies per image.
AI services keep getting better and cheaper with AI-API
access for the leading models dropping in price by 80 or 90
percent last year. Lower prices opened up new use cases and
power small businesses and new market entrants. Open-source AI
models are a particularly powerful vector of competition. Open
models level the playing field, reduce barriers to entry, and
give smaller players access to cutting-edge AI technology
without needing massive resources to train their own models.
This sector is growing fast; Aggregator hugging face, which
hosts more than a million different open source models for
different purposes.
For the most advanced models, open AI just this week, just
yesterday, announced that they will soon release an open-
weights model joining major players like Meta and DeepSeek. The
vibrant AI ecosystem in America does face a threat, however.
Miscalibrated regulation. Regulation often entrenches incumbent
firms by creating barriers that new competitors cannot
overcome, and compliance costs disproportionately burden
smaller newer firms giving established players a competitive
advantage.
The Trump Administration recognizes this threat of
regulation and has shifted the Federal conversation in AI from
one that's based on fear to one that's focused on opportunity.
This is a welcome change.
At the State level, however, there are more than 900 bills
proposing to regulate AI, and they've been just introduced
since January. While often well-intentioned, this emerging
patchwork of regulation threatens to undermine both domestic
competition and our national competitiveness.
Congress should meet this challenge by creating a
consistent national regulatory approach to AI that maximizes
freedom for general purpose AI technology, while clarifying
obligations in specific applications, such as healthcare or
transportation. There are more detailed ideas in my written
testimony.
In conclusion, the U.S. AI ecosystem is remarkably
competitive and innovative, spreading access to powerful tools.
To sustain this vibrant competition and to maintain American
leadership, policymakers should clear regulatory barriers and
resist raising new ones. By doing so, we can preserve the
dynamic nature of AI and continue reaping its enormous economic
and social benefits in the years to come.
Thank you for your attention, and I welcome your questions.
[The prepared statement of Mr. Chilson follows:]
[GRAPHIC(S) NOT AVAILABLE IN TIFF FORMAT]
Mr. Fitzgerald. Thank you, Mr. Chilson. Mr. Coniglio, you
may begin.
STATEMENT OF JOSEPH V. CONIGLIO
Mr. Coniglio. Chair Fitzgerald, Ranking Member Nadler, and
Chair Jordan and esteemed Members of the Subcommittee, I'm
honored to be here today.
In my remarks, I'm going to highlight three key areas where
I believe AI and antitrust overlap, and critique three policy
frameworks for dealing with AI. In my view, AI provides no
reason at this time for either heavy-handed antitrust
enforcement or changing fundamentally our antitrust laws.
First, the AI industry is highly competitive and developing
in a way that is consistent with the healthy innovation
competition that typifies high tech markets. Firms like Nvidia
reinvented themselves with huge success to lead and recoup
their investments in chips, while a number of newer players,
like OpenAI and Anthropic, are flourishing at the model level,
in part, due to partnerships with large digital incumbents.
Indeed, not only is competition robust among cloud
providers, it is a testament to big tech firms not resting on
their laurels, but aggressively also competing at many levels
of the AI stack.
Second, AI is already being heralded as the next general-
purpose technology that, like the internet, will transform our
economy. As the past four years make clear, the stakes for
getting competition policy right are high. Whereas, the U.S.
pursued deregulation and a consumer welfare antitrust model
that was conducive to the innovation and growth that helped
preserve its share of global wealth amidst China's rise, Europe
continued with an antitrust framework more suited to protecting
competitors, missed the digital revolution and saw its share of
global wealth fall by almost half.
Importing a European-style competition policy to America
risks repeating this mistake with AI and ultimately benefiting
China in its own quest for AI dominance.
Third, the question of how AI will affect the antitrust
enterprise itself. For example, when it comes to combating
collusion, some have suggested that AI algorithms may make it
easier for firms to form and police cartel agreements, as well
as facilitate oligopoly outcomes without an agreement and,
thus, avoid antitrust scrutiny.
On the other hand, not only has the Department of Justice
already brought AI-related algorithmic collusion cases using
its existing tools, which in the future, could even be enhanced
by AI technologies that make it easier to detect cartels, but
AI may also increase firms' incentive to disrupt cartel
behavior, resulting in less collusion throughout the economy.
In the face of these complex issues, I want to now discuss
three flawed antitrust enforcement approaches policymakers may
be tempted to adopt.
First, the idea that antitrust is more important than ever
before. Some worry that AI will increase concentration
throughout the economy by virtue of network effects and
increasing returns to scale in ways that make it hard for
markets to self-correct. These concerns, which result in the
antitrust laws themselves, overlook the cost-reducing potential
of AI, which also may lower barriers to entry for small- and
medium-sized businesses and create competition.
Moreover, increased concentration not only isn't bad when
it's driven by lower costs that reduce prices for consumers,
but it often empowers the incentives and abilities that spur
innovation.
Second, there is the contrary view that, because AI is so
disruptive, there's little need to enforce the antitrust laws
at all. Indeed, a similar debate was had during the internet
revolution and noted in the seminal Microsoft case, where the
court discussed how, in technologically dynamic markets,
monopoly power may only be temporary. While, like the internet,
AI is poised to bring tremendous benefits and disruption, the
past 40 years have taught us to be wary of formalizing
theoretical assumptions of market self-correction.
Antitrust still matters, and we should continue to
carefully examine market realities for anticompetitive conduct
in age of AI.
Third, there is the notion that AI may undermine the very
market competition upon which antitrust law is premised. Some
commentators have claimed that AI may even render markets
obsolete through more sophisticated calculation techniques that
could make regulation superior. The central problem with this
perspective is that it's based on a static understanding of how
market competition works.
More dynamic competitive processes will undercut the
justification for State central planning and regulation in a
way that reaffirms the importance of healthy, well-functioning
markets.
In conclusion, I think the right approach is to maintain
antitrust law's traditional focus on promoting competition and
innovation by prescribing collusive and exclusionary
anticompetitive conduct. This has, since the passage of the
Sherman Act, withstood and helped manage the creative
destruction that worked to make America the world leader that
it is today. Ensuring that competition policy remains tethered
to these principles is essential for preserving American
leadership into the 21st century amidst the new wave of
general-purpose technical change that is AI.
Thank you, and I look forward to your questions.
[The prepared statement of Mr. Coniglio follows:]
[GRAPHIC(S) NOT AVAILABLE IN TIFF FORMAT]
Mr. Fitzgerald. Thank you, Mr. Coniglio. Ms. Melugin, you
may begin.
STATEMENT OF JESSICA MELUGIN
Ms. Melugin. Chair Fitzgerald, Ranking Member Nadler, and
distinguished Members of the Committee, thank you for holding
this hearing today and inviting me to participate.
My work focuses on technology and antitrust with the
Competitive Enterprise Institute, a nonpartisan public policy
organization that concentrates on regulatory issues from a free
market perspective. I'm also an antitrust and competition
fellow at the Innovators Network Foundation.
I am pleased to have the rare opportunity to speak to you
today about a bipartisan goal: That of ensuring the United
States remains the leader on artificial intelligence
technologies.
While reasonable people can disagree about the challenges
AI might bring, almost everyone recognizes the technology's
vast potential for economic efficiency gains, life-saving
medical breakthroughs, environmental benefits, and more. There
is similar consensus around the importance of our country
winning the global race for AI dominance ahead of various
authoritarian regimes around the world, including China.
This broad political agreement can be found in last year's
bipartisan House task force report on artificial intelligence
that emphasizes the economic and national security importance
of AI's success for the United States.
If all agree that a flourishing domestic AI industry is
critical, we should next identify the biggest threat to
continued U.S. dominance in the field; namely, government
overreach. Alongside the growing patchwork of State-level AI
regulations that I would encourage this Congress to consider
preempting--but that's a whole other hearing--the threat of
overzealous antitrust regulation looms large.
Default antitrust approach to AI in the U.S. should be one
of humility by regulators. However well-intentioned, antitrust
enforcers are no better than anyone else in foretelling the
future and certainly not clairvoyants capable of ex ante
regulation of a quickly evolving technology.
Antitrust enforcement in the tech sector has been
particularly poor. From an abandoned IBM suit that lasted 13
years to a Microsoft case where innovation in the market
largely rendered the charges moot, litigating antitrust in tech
is especially challenging.
We can look across the pond to the European Union for an
even more cautionary tale in overregulation and excessive
antitrust enforcement. The result is that in the U.S., with its
relatively light-touch regulatory regime and consumer welfare
driven antitrust approach, we boast eight of the top 10 global
tech firms by market cap, with Europe unable to claim even one.
Currently, the EU has an AI regulatory framework much
larger than its AI industry, putting them on the now familiar
path to diminished prosperity. We needn't make the same mistake
here. We know what regulatory framework leads to success and
can course correct for the harmful antitrust trends of the last
four years.
The embrace of a big-is-bad approach to antitrust
enforcement at the prior administration's Federal Trade
Commission and Department of Justice must be reversed. It's
critical that America's AI ecosystem has both large established
firms with their vast resources for research and development,
and vibrant small startups with their nimbleness and
innovation.
In many cases, startups rely on the expectation of
eventually being profitably acquired by large firms to secure
early funding. This is why continued regulatory hostility to
mergers and acquisitions is a recipe for stifled AI growth.
We must also return to an economics-based approach to
identifying competitive problems should they arise. Whatever
challenges outside of measurable economic concerns AI might
bring, antitrust law will be the wrong tool for correcting
them. Refocusing enforcement on consumer welfare and away from
ambiguous vague social goals will be a productive improvement
over the last several years.
What are the current facts on the ground for AI competition
landscape right now? So far, competition seems to be thriving
in what is a nonunitary market from hardware to the data layer
to algorithms and applications.
Then newcomer OpenAI had a meteoric rise to popularity with
its debut of ChatGPT in November 2022, and is holding its lead
over even the biggest established tech competitors as the most-
used large language model. Those tech giants plan to spend more
than $300 billion this year investing in AI technologies, which
is not the behavior one would expect from entrenched
monopolists insulated from competitive pressures.
The presence of multiple open source models that allow the
little guys to build on those already established models
ensures beneficial competitive pressure and leaves space for
innovation.
The current State of competition in AI is strong. The
threat of premature and unwarranted antitrust intervention
poses one of the greatest threats to global U.S.
competitiveness.
Thank you for this opportunity to testify, and I look
forward to your questions.
[The prepared statement of Ms. Melugin follows:]
[GRAPHIC(S) NOT AVAILABLE IN TIFF FORMAT]
Mr. Fitzgerald. Thank you, Ms. Melugin. Mr. Bedoya, you may
begin.
STATEMENT OF ALVARO M. BEDOYA
Mr. Bedoya. Thank you, Chair Fitzgerald, Ranking Member
Nadler, Chair Jordan, and the Members of the Subcommittee.
Thank you for having me here.
Right now, people think that a successful AI startup needs
four things: Chips, cloud, training data, and the people to
train those models and deploy them. So, right now, there are
races being run to secure access to those four key inputs.
In the time I have with you today, I want to issue three
warnings about those races. First, I want to warn that certain
agreements between major AI developers and two of the biggest
cloud service providers may be limiting the availability of
those cloud providers to those smaller innovative startups.
Just this January, FTC issued a staff report. We looked at
three partnerships between the major cloud providers and the
major AI developers out there. The report showed that cloud
providers typically got things you expect, equity, a share in
revenue from those AI developers. The agreements also required
the AI developers to spend a lot of the investment from the
cloud provider on cloud services from the cloud provider.
Now, unfortunately, these kinds of terms may be
incentivizing those cloud providers to limit access to their
platforms away from AI startups who they're not partnered with.
Unfortunately, this is not hypothetical. There were actually--
there was actually an internal document from one of those cloud
providers that said, quote,
We are unable to serve a growing number of startups who need
bursts of graphics processing unit capacity to train and
release their models.
So, I agree, we want a world where a scrappy startup can
eat a powerful incumbent. We have to ask if these kind of
agreements, this kind of vertical integration, is leading us
away from that kind of innovation. That's the first warning.
I also want to address the race to collect and retain data
because in the race to scrape by access, all kinds of human-
generated data, I'm afraid that a lot of AI developers are
collecting and keeping data that a lot of us would consider
deeply private.
So, here's an example. In 2023, the FTC alleged that, by
default, Alexa was keeping recordings of kids' voices forever,
and it wasn't just that. Even when parents said, ``Hey, Alexa,
I want you to delete that,'' Alexa wasn't doing that; it was
actually keeping the transcripts of that information.
Now, anyone who is a parent should stop and ask themselves,
what on Earth does my kid need to say to a smart speaker for me
to stop everything and say, Oh, no, we need to delete that
data. That's not what happened.
Here's a key thing, though. The FTC alleged that one of the
reasons Alexa--Amazon gave for retaining that data was their
need to train their voice recognition algorithms. So, that's my
second warning. We have to remind companies that the need to
train your algorithm doesn't override the need to follow the
law, including privacy law and kids' privacy law.
Here's the third and last flag. There's been a lot of
attention--and I think for good reason--paid to the apparent
collection and use of copyrighted works to train models. My
colleague, Mr. Chilson, mentioned the Studio Ghibli algorithm
that's out. I'm afraid we aren't paying enough attention to the
collection and training--of training data about people who are
in no way famous. Working people. Let me give you a quick
example.
During the actors' strike, a prominent director, Justine
Bateman, posted online asking if any extras--these are folks
known as background actors--if any of them had ever been asked
to submit to body scans on set. Suddenly these stories started
pouring in, people saying, yes, I got tapped on the shoulder, I
was told to go to this tent, and I stepped in there, and I was
scanned 360 degrees, head to toe. I asked why; they didn't tell
me. I asked if I was going to get paid; they didn't give me
anything. The few people who found out were told, Oh, yes, this
is to add you in later if we want to. Of course, at no
additional pay.
I'm talking about this because scanning people at their
jobs to replace them at those jobs is becoming common outside
of Hollywood. So, this is starting to happen in customer
service call centers. These are jobs that used to provide high
school graduates with a path to the middle class, but now
they're rife with surveillance. A lot of that surveillance
appears to be aimed at recording the best performers to create
algorithms that are marketed as replacements for the workers at
those call centers.
So, in closing, I would respectfully urge the Subcommittee
to not just look at competition between AIs, but competition
between AIs and workers, because it raises some pretty
fundamental questions like how do you compete with a digital
version of yourself who can work 24 hours a day seven days a
week with no break, no vacation, and no benefits? Is it fair to
force you to train your AI replacement? Is it fair to force you
to do it without your knowledge?
As I explained in my testimony, I think this is a great
reason to pass a workplace privacy law.
Thank you again, Mr. Chair, Ranking Member Nadler for the
opportunity to testify. I'm glad to answer any questions you
may have.
[The prepared statement of Mr. Bedoya follows:]
[GRAPHIC(S) NOT AVAILABLE IN TIFF FORMAT]
Mr. Fitzgerald. Thank you very much, Mr. Bedoya.
I'll now proceed under the five-minute rule with questions.
I'm going to recognize myself.
In recent remarks at the AI Summit in Paris, Vice President
JD Vance stated that AI presents the extraordinary prospect of
a new industrial revolution, but it will never come to pass if
overregulation deters innovators from taking the risks
necessary to advance the ball. He was referring to the European
Union's Digital Markets Act, and AI Act.
Ms. Melugin, why would replicating Europe's heavy-handed
regulation of AI in the U.S. certainly be maybe unwarranted, or
even premature?
Ms. Melugin. It's kind of the European Union to have run
this experiment for us. It has not gone well. Their first big
attempt at digital regulation, the GDPR ended up hurting the
entire industry, but mostly hurting the little guy
disproportionately, which we would see the same here with a lot
of the regulatory moves. The big entrenched accused
monopolists, or just market leaders, depending on your
perspective, might have the resources to comply with a lot of
those regulations, but the runway is not clear for the next big
success story.
That's what you have to keep in mind, that a lot of times,
the companies that have a seat at the table might be able to
make this work, but we don't know what's being developed in
someone's garage right now that can be the next big thing. That
goes to the general approach in Europe, which has been that if
you're chilled in your innovation by worrying that everything
you do, every new business arrangement, every new idea could be
running afoul of one of the many--the DMA, the DSA, and the
GDPR, it just has an overall chilling effect on that industry,
and, of course, relatedly, on investment in that industry.
Frankly, the U.S. really doesn't have that time. If China--
I don't think it's an overstatement to say they're nipping at
our heels in terms of the AI race. What those companies and
those industries are doing in China is not the regulatory
framework you see in the U.S., but it's also very much not the
one you see in the EU. It's no coincidence that you find all
three of those segments in a different order of success. We
want to make sure we keep ourselves out front.
As I said in my testimony, the European Union doesn't have
one of the top 10 global businesses. Again, that's not to say
that the small companies aren't important, too, to develop here
in the U.S. You certainly need the scaling abilities of large
companies, the research and development resources of these
large companies. You want a healthy ecosystem with the big guys
and the small guys. Adopting a European approach won't allow
that to happen at either level here in the U.S.
Mr. Fitzgerald. The Netherlands-based software firm, Bird,
which has been recognized kind of globally at this point, is
moving out of the EU as a direct result of the AI Act. What
they said was that Europe regulators are just killing
innovation within the EU.
You think there's lessons already that can be learned here
in the States from what we've witnessed on some of these
examples of what's transpiring overseas?
Ms. Melugin. Yes. I don't even think you have to abstract
very much. The EU AI Act that those regulators bragged about
being the first to regulate AI--which to some of our ears,
doesn't sound like something to brag about--means that a lot of
the U.S. companies that rolled out AI products that consumers
have found incredibly useful, have been very popular here in
the U.S., they were delayed in Europe because they weren't sure
that they could comply with all the rules and regulations.
At the worst-case scenario, you don't get those
innovations, constituents don't get those services and those
tools. The best-case scenario is that they're delayed because
of regulatory overview. Again, that's an example where the
little guy who is innovating something new maybe doesn't have
the team of lawyers that the big guys do. I don't think we want
to set up that dynamic here in the U.S.
Mr. Fitzgerald. Very good. Mr. Coniglio, you just cautioned
against overregulation in your testimony. You note that robust
antitrust enforcement promote customer welfare and innovation
to remain a policy focus. Can you just elaborate kind of
further on that point?
Mr. Coniglio. Absolutely. Thank you for the question.
Ms. Melugin pointed out the lessons we learned from
regulation. I will note, the U.S. also has some lessons that it
learned from deregulation about how deregulation can spur
economic growth and innovation. There's also a very important
story to tell regarding competition policy, and specifically,
two different approaches that were taken over the past 40 years
between America and Europe on competition policy.
At a high level, Europe's approach was very much focused on
protecting competitors. It wanted to see markets where there
were a lot of small firms. America's approach was focused on
not condemning size in and of itself but focusing on consumer
welfare and innovation.
If you look most recently at the Draghi report that was
issued in Europe, it was very clear that the digital revolution
that happened in America has really driven the productivity gap
between the United States and Europe that's occurred. Antitrust
is not the sole reason for that, but I think it's part of the
reason. Because if you have an antitrust policy that's not
conducive to innovation and consumer welfare, and instead,
focusing on protecting small competitors, you're not going to
see the dynamic innovation competition that's really driven the
digital industry in America, and which, in part, large firms
can also play a big role in driving that innovation.
Mr. Fitzgerald. Thank you very much. My time has expired.
I'm going to recognize the Ranking Member, Mr. Nadler, for
five minutes.
Mr. Nadler. Thank you, Mr. Chair.
Commissioner Bedoya, one of the many areas in which the
Federal Trade Commission has distinguished itself in its
groundbreak-
ing activity to combat anticompetitive practices in the
pharmaceutical industry. Whether it's the pay-for-delay cases
or efforts to fight misuse of the FDA publication called the
Orange Book, the FTC has been at the cutting-edge.
That's why I was so disturbed to see that yesterday, the
FTC's general counsel stayed the administrative adjudication of
the commissioner's case against the three largest pharmacy
benefit managers over the price of insulin. The effort to bring
down prices for consumers is just one example of the work you
were doing before President Trump attempted to remove you from
office.
Can you explain the costs to be borne by consumers
associated with the President's unlawful attempt to fire you
and Commissioner Slaughter?
Mr. Bedoya. Thank you, Ranking Member Nadler. Maybe I can
just talk about this case. I'm pretty sure everyone in the room
cares about the cost of insulin. This is a case where FTC staff
is alleging that these pharmacy middlemen--they're called PBMs,
pharmacy benefit managers--are competing not to lower the price
of insulin, which is how markets normally should work, they
allege instead that these pharmacy middlemen are competing to
raise the price of insulin.
Commissioner Slaughter, who I'm joined--
Mr. Nadler. Competing or colluding?
Mr. Bedoya. Pardon me? I didn't hear the first part--
Mr. Nadler. Competing or colluding?
Mr. Bedoya. What was the first--beating or colluding?
Mr. Nadler. Competing or colluding?
Mr. Bedoya. Competing. The allegation in the complaint is
there's competition to raise the prices.
I'm joined by Commissioner Slaughter. She and I are the
only judges who are presiding over that case in an appellate
manner because our colleagues, Chair Ferguson and Commissioner
Holyoak, are both recused.
Now that it's the administration's contention that we are
removed, there are no judges to adjudicate that case. The
pharmacy middlemen have moved to stay it. That move--that stay
was granted--that request for stay was granted yesterday, and
this case won't be adjudicated for at least another 105 days.
That's 3\1/2\ months. The evidentiary hearing, I understand,
will occur even later.
Basically, it's permanent--not permanently, but
indefinitely on ice, and I don't think anybody wins from having
a case like that not adjudicated.
Mr. Nadler. Thank you. We know that, unfortunately, one of
the ways new technologies is deployed is to find ways to break
the law. Is computer-generated pricing or algorithmic pricing
being deployed to violate the antitrust laws, and what can the
FTC do about it?
Mr. Bedoya. One of the things we're hearing a lot are
allegations that, instead of outright colluding--to use your
term--just sitting down in a smoky room like the Beef Trust did
in Chicago in the early 1900s and just set prices.
What a lot of competitors are doing is saying, Oh, no, no,
we compete, but we happen to use the same algorithmic pricing
software. Although there are allegations, for example, in the
RealPage lawsuit from DOJ that they are, in fact, also picking
up the phone and just straight up colluding, increasingly what
you see are cases where they just use the same software. This
is absolutely a core concern.
The FTC has issued various requests for information around
aspects of this. We issued an interim report around algorithmic
pricing in January, and it's something we need to continue to
track closely.
Mr. Nadler. Thank you. We all recognize that AI carries
with an enormous potential, as well as possible perils. We also
all recognize that it's a field that continues to develop often
at a rapid clip.
Can you explain why the FTC's involvement in evolving
industries, even while they're undergoing changes, is a good
thing?
Mr. Bedoya. Absolutely. I understand folks who say, ``You
don't want too aggressive antitrust enforcement, that's a bad
thing.'' For me, the case that jumps out is the case you
mentioned, the proposed Nvidia-Arm merger. Nvidia is a chip
supplier, Arm is a chip designer; it's a vertical deal. Arm was
valued at the time of that deal at $40 billion. FTC sued. They
dropped that merger. As of 9:58 this morning, Arm was valued at
$114 billion.
The answer to your question, Congressman Nadler, is that's
so the market can continue to evolve; so that scrappy startup
can access all of these critical resources that the companies
like these provide without the limitations like the ones I
referenced in my opening statement where the incumbents are
getting locked in and startups are getting squeezed out.
Mr. Nadler. Thank you.
Finally, fields of law--all fields of law can and should
evolve over time. In recent years, there's been increased
attention as to how competition affects not only consumers, but
also workers.
With that perspective in mind, how should we be thinking
about how AI might impact the American workforce?
Mr. Bedoya. I would say the biggest thing people are not
talking about is the need for some kind of protection for
workers in the workplace so that they just can't be scanned
head to toe and replaced by algorithms. This isn't no longer a
thing from Hollywood. This is something you're seeing in jobs
like call centers, online therapy, et cetera.
Mr. Fitzgerald. Gentleman's time has--
Mr. Nadler. My time has expired. I yield back.
Mr. Fitzgerald. I now recognize the gentleman from Texas,
Mr. Gooden.
Mr. Gooden. Thank you. Ms. Melugin, the Biden
Administration passed sweeping Executive Orders on AI which
weren't authorized by any congressional action or directive.
Do you think the previous administration bypassed Congress
and micromanaged AI development without public accountability?
Ms. Melugin. Yes.
Mr. Gooden. Would you like to comment further?
Ms. Melugin. Sure. I do. I think that it's a perfect
example of the rush to regulate. What you are doing here today
is having a very constructive conversation about one aspect of
AI, and that's to be applauded.
Coming out with an EO that puts large restrictions on how
people are allowed to innovate, testing, rules that haven't
really been put through either the common-law process in the
courts or a congressional process is always a mistake. Even if
I had agreed with the rules that were there, I wouldn't support
it. I didn't. The lifting of that executive order with the
current administration is generally a very positive step.
Mr. Gooden. What role should Congress play, in your mind,
in ensuring that AI governance reflects competitive neutrality,
market dynamics, and constitutional accountability?
Ms. Melugin. As I somewhat snarkily mentioned in my
testimony, I do think that the biggest thing Congress could do
would be to consider preempting the rush of State regulations
on AI. I think there's very well-intentioned State legislators.
They see that Congress isn't moving on this. They want to do
something.
A patchwork of potentially 50-plus different regulatory
regimes for AI is the kind of hurdle that could really slow
U.S. innovators down, and certainly not the kind of hurdle that
they're facing in China. It's an infinitely, like,
unpolitically satisfying answer, but I think Federal preemption
should be something serious lawmakers consider.
To the antitrust question that we're talking about today,
let's wait and see until what problems arise, and also, what
laws and tools are already on the books that regulators and
lawmakers can use to react to those problems. Price fixing is
still price fixing, whether it's done in a smoky room with
meat--which actually sounds kind of great, the meat part.
Mr. Bedoya. The meat wasn't--
Ms. Melugin. The meat wasn't--OK. It was a good visual,
though. I want to give you full credit for that--or done with
AI technology in an algorithm. These things are still illegal.
Fraud is still illegal whether it's--whatever kind of
technology you use.
Taking things on a case-by-case basis is a smarter approach
to not making the mistake of overregulating and hurting
innovation in the U.S.
Mr. Gooden. You mentioned China. Do you feel--where do we
rank in terms of attracting investors, and are we doing better
or worse than China with respect to AI?
Ms. Melugin. Well, I think the structure is so different,
right? Happily, it's private investment here in the U.S. for
the most part. China is pouring billions of dollars into these
technologies, and I do think that without tying one arm behind
our back, my money is on the U.S. I would innovate always.
I do think that this State patchwork of regulations or
unwarranted and overzealous antitrust intervention could put us
at a real disadvantage. We've been saying, there's different
points along the AI ecosystem. It's hard to give you one answer
about all of it, but it's a close race.
Mr. Gooden. Thank you. Yield back.
Mr. Fitzgerald. Gentleman yields back.
Gentleman from California is now recognized for five
minutes.
Mr. Correa. Thank you, Mr. Chair. I want to welcome our
witnesses today.
Every day in the newspaper, something new with AI, new
developments, new companies, competition, innovation,
medicines, new medicines, drug developments, manufacturing, and
supply chains. A lot of good stuff. I do believe we need
guardrails, but they have to be guardrails that don't harm
innovation or harm American firms.
The U.S. right now, we, the United States, still lead the
world, used to be by a couple of years. Now, it's maybe two
months, maybe couple of weeks, compared to China.
Mr. Chilson and Mr. Coniglio, if I can ask each one of you
a yes-or-no question. Gentlemen, would you say that university
research played an enormous role in developing AI? Yes, no?
Mr. Chilson. Yes, over the 40 years.
Mr. Correa. Mr. Coniglio?
Mr. Coniglio. Yes, I would say the same.
Mr. Correa. If you look at this chart, you can see that it
is, in fact, true. Private companies are developing and
investing in AI.
Second question, gentlemen: The future workforce, training
that AI workforce, research still happening at universities,
correct?
Mr. Chilson. I think some of it. Lots of companies--
Mr. Correa. Some of it. OK. Mr. Coniglio?
Mr. Coniglio. I would agree.
Mr. Correa. If we cut out funding for university research
today, are we damaging the American pipeline, students--future
students who become leaders in AI, fundamental research, are we
hurting ourselves by cutting off university research funding?
Mr. Chilson, Mr. Coniglio? Yes, no?
Mr. Chilson. I'm honestly not sure.
Mr. Correa. Mr. Coniglio?
Mr. Coniglio. I would certainly say that funding from
universities is very important to drive innovation. That's true
generally, yes.
Mr. Correa. I say that in the context of our main
competitor, which is China, which are essentially doubling down
their investment in this area. That's well-established; that's
well-known.
Now, let's look at this chart of some of the top AI
companies. Data from 2023. Many of these firms are started by
immigrants. This is in 2023. Can you imagine the number of
immigrants started AI firms for 2024? Probably a whole lot
more. This doesn't even reflect major tech started by Sergey
Brin of Google or Elon Musk.
So, gentlemen, another question for you. Do you agree that
immigrants have played a huge role in developing and advancing
technology in AI sectors? Yes, no?
Mr. Chilson. Absolutely.
Mr. Correa. Mr. Coniglio?
Mr. Coniglio. Yes.
Mr. Correa. This country--we really want immigrants in this
country who can contribute to innovation, and we don't want to
scare them away, have them go to Canada, China, India, not even
come to this country. Would you say that's--yes, no, Mr.
Chilson?
Mr. Chilson. America is the best place to do AI in the
world.
Mr. Correa. Mr. Coniglio?
Mr. Coniglio. I would agree we need to keep America an AI
leader and use all tools we can to make that happen.
Mr. Correa. Over the last 50--more like 100-200 years,
America has always attracted the best and brightest from around
the world because they know opportunities in this country
exist.
I agree that overregulation, misplacement of antitrust
enforcement can harm AI, can harm this country's companies as
they seek to compete with China. I'll tell you, looking at this
administration's policies of cutting research funding to
universities, scaring immigrants away from studying or creating
new companies in the U.S., creation of unstable markets through
tariffs, these three factors, in my opinion, are going to hurt
this country even more; long-term damage, long-term damage to
our leadership, and our economy in the future. I do hope we
begin to focus on what's made this country great, what helped
us in the last 50 years innovating, leading the world because
we can't afford to lose the AI race.
Thank you very much, Mr. Chair, and I yield.
Mr. Fitzgerald. Gentleman yields.
Mr. Correa. Can I add some articles for the record?
Mr. Fitzgerald. Gentleman is recognized.
Mr. Correa. Chair, I'd like to ask unanimous consent to
introduce the following articles:
(1) ``China's top universities expand enrollment to beef
up capabilities in AI.''
(2) ``65 percent of top AI companies have immigrant
founders.''
(3) ``Trump's science policies pose long-term risks,''
New York Times, March 31st.
(4) ``Trump administration has begun a war on science
researchers,'' New York Times, March 31st.
(5) ``China leads the world in generative AI adoption.''
(6) ``Former Google employees create 14 of the top 50 AI
startups.''
(7) ``Trump loves AI, but his tariffs ramp up costs for
data centers.''
(8) ``Apple chip engineers return to China. Joins Fudan
University amid push for talent.''
(9) ``Immigrant entrepreneurs bring jobs and
innovation,'' Forbes, May 23, 2024.
(10) ``From underdog to genius, Yale computer scientist,
Sun Huanbo, returns to China.''
(11) ``Tech companies are telling immigrant employees on
visas not to leave the U.S.''
Thank you, Mr. Chair.
Mr. Fitzgerald. Without objection.
Mr. Fitzgerald. The gentlewoman from Wyoming is now
recognized for five minutes.
Ms. Hageman. Thank you. China is trying to control the
global AI sector through extensive state-directed investments,
aggressive subsidies, and protectionist trade measures. It is
making substantial investments to secure its global position
through initiatives like Made in China 2025 and the New
Generation Artificial Intelligence Development Plan.
Mr. Chilson, how might China's government subsidies and
protectionist policies affect global competition in AI?
Mr. Chilson. China's model for AI innovation is very
different than the U.S. It's largely State-sponsored. It is
driven through a structure that is designed to make sure that
the Chinese Communist Party has its finger in everything, and
it means that the end result is designed from the top
ecosystem. They pour a lot of money into this, and they have
talented people, and so they do have surprises occasionally.
The deep seek AI model that came out, they can get a lot done
with that model.
Ultimately, however, the U.S. model, if we don't tie one or
both hands behind our back, is going to win in this space,
because we have the talent, we have the investment, and we have
the ecosystem that rewards trying new things out and seeing if
they work. I worry a lot about China. I think modeling our--and
this is why I was a little bit confused about the university
question. A top-down investment model that tries to look like
China is not the right model for the U.S.
Ms. Hageman. I'm glad to hear that, and I'm also excited to
hear about your optimism, because what would be the strategic
consequences if China achieved dominance in the global AI
space?
Mr. Chilson. Well, there's a lot of different consequences.
We can just look at cultural ones. We can look at strategic
national security ones, and those are the ones that raise the
most eyebrows.
Ms. Hageman. Can you give me some examples?
Mr. Chilson. Yes. This technology is very powerful, both in
an offensive and defensive manner for cybersecurity, so both to
defend from cybersecurity attacks and to cause them, and so,
being ahead in that race matters a lot. A lot of our
infrastructure in the U.S. obviously is--needs to be strong
against those types of attacks, and I think that AI has a huge
potential to be a defensive barrier to those types of things,
but we need to stay ahead in the race.
Ms. Hageman. Well, I agree with you on that as well. This
Committee has observed with concern harmful policies also
coming from the EU on privacy, antitrust censorship, and more.
Now, the Artificial Intelligence Act will impose significant
compliance costs on affected companies, particularly American
businesses that play a pivotal role within the AI sector.
Mr. Coniglio, how should we, as Members of Congress, be
concerned about the impact of this framework on our constituent
businesses, and are there specific ways in which the EU
regulations might disadvantage smaller American AI companies as
compared to larger firms?
Mr. Coniglio. Thank you very much for the question,
Congresswoman. I would say we should be very concerned about
the regulatory approach that the European Union is taking, not
just with the AI Act. One of the things I would say they did
learn from the GDPR is now to do regulation that targets
American firms, specifically, so that they don't burden their
smaller firms with regulation. That's really been the lesson
with the DMA, which is absolutely disproportionately having
very harmful effects on American businesses, really our tech
leaders, and it was intended to do so. As I recall a letter
from Chair Jordan, starting to investigate those practices is
very important and something we need to continue to do.
Ms. Hageman. Do you have recommendations as to what
Congress specifically ought to do in this space?
Mr. Coniglio. It's a very complicated area. I would note
with respect to potential legislation; in Korea, there was a
proposed bill by Representative Miller, I believe, to talk
about potential tools in a trade area that could be used to the
extent that regulations are being basically weaponized as de
facto protectionist policies against American companies, and I
think that might absolutely be the right framework to think
about it.
Ms. Hageman. OK. Mr. Chilson, in your testimony, you cite
how competition is robust and global across all layers of the
AI stack. How should Congress consider protecting American
companies in such an environment, or is such a policy even
achievable or worthy of consideration?
Mr. Chilson. American companies lead the world in this, so
I don't know that they need protection from Congress, but what
they do need is the ability to continue to innovate, build, get
the supplies and the data that they need, and there's a lot in
the regulatory space that Congress could work on. In
particular, we've mentioned it a couple times, but one of the
biggest threats is well-intentioned, but often very abstract
and not well-developed, state-level regulation, which could
limit a lot of this innovation.
Ms. Hageman. So, preemption is important?
Mr. Coniglio. I think it's very important.
Ms. Hageman. Thank you, and I yield back.
Mr. Fitzgerald. Gentlewoman's time has expired and now--
Mr. Nadler. Mr. Chair, I have a unanimous consent request.
Mr. Fitzgerald. Gentleman is recognized.
Mr. Nadler. Thank you. Mr. Chair, I'd like to enter in the
record an article ``AI Watch Global Regulatory Tracker'' that
details how many more regulations than the U.S. does on AI
development and deployment.
Mr. Fitzgerald. Without objection.
Mr. Fitzgerald. The gentlewoman from Vermont is now
recognized for five minutes.
Ms. Balint. Thank you, Mr. Chair.
Commissioner Bedoya, thank you so much for being here, and
Commissioner Slaughter, it's really nice to see you.
Commissioner Bedoya, as commissioner at the FTC, your
experience of competition, artificial intelligence, and how
these things affect everyday people is extremely important to
this discussion. We all understand that we're in the early days
of what may be a massively transformational technological
moment. It's plausible that this may be like the emergence of
the internet, the mobile phone, or really even electricity.
It's so important that we don't lose sight of the central point
here, which is how does this technology help or hurt working
people? That's what I'm focused on.
Commissioner Bedoya, you spent a great deal of time as FTC
commissioner outside of D.C. traveling around the country,
hearing from people, and I want to know, have you heard a sense
of concern from people, working people, about how AI is
impacting their lives?
Mr. Bedoya. Sure. I confess a lot of my roundtables are
focused on groceries, pharmacy, and that kind of thing.
Ms. Balint. Everyday costs.
Mr. Bedoya. What's that?
Ms. Balint. Everyday costs.
Mr. Bedoya. Everyday costs, that's right. I will say a
couple things on this. A lot of people look at what happened
around the actors and writers strikes and they say, ``Oh,
that's just a Hollywood thing,'' but entry level actors and
entry level writers don't make that much money, and you had
situations where people were quite worried, and with good
reason, that the cost of selling your first script was going to
be to submit all your work to be fed into one of these AIs so
that there would be no need to hire you in the future.
I also do think there have been a lot of concerns, my
colleague Commissioner slaughter has done a roundtable on this,
on the use of price fixing algorithms in rental housing
markets.
Ms. Balint. That's where I'm going.
Mr. Bedoya. Yes. That's a real concern.
Ms. Balint. That's exactly where I'm going. Let's jump to
that, because this is a deep concern of mine. Could you tell us
in real terms, because you had talked about earlier that it's
not just the algorithm itself in terms of real page, but there
seems to be instances that we can look at, there seems to be
evidence that there is actually some collusion happening here.
That's something that I'm deeply concerned about, and we all
should. There is a crisis of housing in this country and
renters are over a barrel right now. They don't have a lot of
options. Can you talk a little bit about how this is impacting
regular people just in terms of trying to afford the rental
housing.
Mr. Bedoya. Sure. The allegations in this space are that
these algorithms don't just offer the exact right amount of an
increase so that a renter swallows the increase rather than
tries to find a different place. There're allegations of actual
overt in the RealPage lawsuit calling between landlords to
coordinate, but there's also allegedly information sharing
creating a pool of data that these landlords can draw on to
increase rent without saying--while being able to say, Oh, we
never called each other.
One really interesting point here is sometimes antitrust
harms, because they're a little complex, can get a little
technical. I'll just point out that one of the corporate
landlords in the RealPage suit is a company called GreyStar.
FTC is also simultaneously suing GreyStar for packing people's
rent checks with these fake hidden fees, like, Oh, you want no
cockroaches in your apartment? You've got to pay extra for
that.
Ms. Balint. Pay extra for no cockroaches, huh?
Mr. Bedoya. Precisely. You see anticompetitive conduct arm
in arm with conduct that hurts consumers, in my view, pretty
obviously.
Ms. Balint. I do think it's important in this hearing as we
talk about the importance of not stifling innovation to, at the
same time, be able to hold both truths. We can be careful in
our regulations. We can be careful in the guardrails that we're
putting on, but I don't think anyone here, at least I hope not,
that anyone on the Antitrust Committee would be supportive of
landlords colluding to raise the price of rent, and I know in
some instances, we've seen that this AI dynamic pricing scheme
has led to increases of over $800 over the course of a year of
not just these hidden fees that we're talking about, but the
prices that they're paying for rent itself.
Of course, you've talked about how essential workers are
being replaced in key industries, where we see patients
illegally being denied care because of an AI-based
recommendation, where artists are being robbed of their work. I
just have to say, we have got to remember the importance of
following the law when it comes to having FTC commissioners,
and I think it's outrageous that you and Commissioner Slaughter
were illegally removed from your positions. We need you on the
watch to take care of my renters back home in Vermont who are
being screwed over.
Mr. Fitzgerald. Gentlewoman's time is expired.
Ms. Balint. I yield back.
Mr. Fitzgerald. Gentlewoman yields back. I now recognize
the gentleman from North Carolina, Mr. Harris, for five
minutes.
Mr. Harris. Thank you, Mr. Chair, and thanks to all of you
on the panel for being a part of this hearing today. On January
23, 2025, President Trump issued an Executive Order titled,
``Removing Barriers to American Leadership and Artificial
Intelligence,'' aimed at solidifying the United States'
position as the global leader in AI. I, for one, applaud the
Trump Administration's leadership on this issue.
Mr. Chilson, would you just take a moment and explain how
this Executive Order will indeed help solidify our position as
the global leader in AI as the EO states?
Mr. Chilson. Well, it's a sea change from the Biden
Executive Order, which was the third longest Executive Order in
history, and basically set a governmentwide call for agencies
to figure out how to regulate AI. This Trump Executive Order is
the opposite. It's focused on figuring out how we can create
the opportunities, or how can we maximize the opportunities
from this amazing technology which is grown here in America.
Some of its mechanisms. It directs the Office of Science
and Technology Planning to create an AI action plan. Those
comments have gone into that agency, and they're going to come
out with a comprehensive approach.
The main thing here is the total change in perspective.
It's gone from fear and risk that this technology is going to
hurt people to recognizing that it has huge opportunities, not
just for consumers, not just for everyday price--everyday
things that they use, but also, in creating a lot of
competition in the AI industry, but also across the economy as
a whole.
Mr. Harris. You reference the science and technology
policy. I know Section 4 of Trump's Executive Order requires
various executive agencies to coordinate on the creation of an
action plan to submit to the President in 180 days, and the
action plan should lay out how we can achieve AI dominance. You
and your colleague, Mr. Chilson, I understand that the
Abundance Institute submitted recommendations to the Office of
Science and Technology Policy on what should be included in
that action plan. Could you take just a moment and summarize
what your recommendations might be?
Mr. Chilson. Right. At the very top, you've heard this
concern repeatedly across this hearing is stemming the tide,
the flowing tide of State AI regulations. What we have right
now is a set of regulatory schemes that regulate from the very
top, the most abstract level of AI, these general-purpose
models. It's a general-purpose technology that can be used in
millions of different ways, and States are trying to say that
these companies have to make sure that this model can never be
used in a way that might hurt somebody, and that's the type of
regulation that is not only impossible to comply with, it will
be impossible to enforce as well.
Congress needs to look at this--the White House needs to
look at different ways to stem this tide, whether through the
bully pulpit or proposing legislation to Congress that will
help stem that flow. I think it's a big challenge in Congress,
and the White House needs to step up to that.
Mr. Harris. Well, let me ask you this: Can Congress play a
role in helping the United States achieve the goals that have
been set forth in the Executive Order?
Mr. Chilson. Absolutely. It will be essential that Congress
does so.
Mr. Harris. OK. Are there any legislative efforts that will
bring us toward the vision set forth by the Trump
Administration?
Mr. Chilson. Well, I think there's a lot of different
models we can take. One State action that has been very useful,
Utah has an AI act that creates an AI learning laboratory where
people can come in, get regulatory mitigation when they don't
understand fully what the risks are because their technology is
new, and both the legislators and the company can learn
together.
The most recent example of this was a technology that
creates--it's an AI mental wellness app for teenagers, a really
important problem. You can understand the sensitivity there.
They worked with this regulatory sandbox to create not only
just the app itself, but also there's been legislation that was
proposed to Utah to the legislature that governed some of those
types of apps, and so it's a cooperative thing. It's the kind
of structure that might work very well at the Federal level as
well.
Mr. Harris. Very good. Thank you so much.
Mr. Chair, I yield back.
Mr. Fitzgerald. Gentleman yields back. The gentleman from
Illinois is now recognized for five minutes.
Mr. Garcia. Thank you, Mr. Chair, and to all the witnesses
here today, to Commissioner Slaughter for also being here to
observe the proceedings.
As usual, so far in this hearing, it looks like my
colleagues, the Republican side of this panel, is here to
champion corporate interest over workers and consumers. Today,
their obedience is making them argue against even reasonable
regulation of the AI industry. While not surprising, it's still
incredible to see them making these claims with a straight face
given everything that's going on.
Let's start with the fact that the Trump Administration and
its billionaire overlord, Elon Musk, are weaponizing AI to
carry out their lawless and unpopular agenda. They're using AI
to arbitrarily surveil students, expand the power of DHS and
ICE and gut Federal agencies. In the hands of billionaires and
authoritarians, like Musk and Trump, AI is a loaded gun waiting
to be used on immigrants, workers, and consumers. This is
exactly why it's absurd to oppose commonsense regulation of the
AI industry.
Let's talk about what that commonsense could be. It's also
absurd, because we know what happens when we don't enforce
antitrust and consumer protection laws against technology
companies. As my colleagues have pointed out, the failure to
apply these laws to big tech companies at the dawn of the
social media age has led to the dominance of these companies,
which have routinely abused workers, exploited consumers,
deployed their technology to expand surveillance, and trample
on privacy rights. That is the consequence of inaction. We must
learn that from our mistakes and we must prioritize workers and
consumers over corporate profits.
Let's talk about what that looks like. Commissioner Bedoya,
thank you for being here today. I stand in solidarity with you
as you face Donald Trump's illegal efforts to fire you and
Commissioner Slaughter. Can you talk about the competition
between AI, if it exists, and workers and the implications for
labor rights?
Mr. Bedoya. Sure. I was down in Bayou La Batre, Alabama, a
couple months ago, and there was a case from Pascagoula, which
is right near there, of a call center worker who had a high
school diploma, a single mom, and she had been able to take
this job and find the middle class. She bought a house, nice
neighborhood, great schools. What was starting to happen is
that there was this AI transcript that was being generated of
her calls, but it was wrong.
People on the Gulf, my wife from Louisiana, have a
different cadence, different accent slightly, and the AI-
generated transcript was failing, and she would get docked for
those errors. One of the things she openly wondered was--
because there was also other software operating that the
workers thought was going to be used to replace them. She
openly wondered, ``am I training my replacement?'' These are
the kind of folks who may be hurt by these efforts to scan,
track, surveil, and swallow up all this data about people
working really important jobs that might be used to replace
them.
Mr. Garcia. Thank you for sharing that, and from basically
on your work at the Federal Trade Commission, what kinds of
protection do you think are needed for workers if Congress were
to act?
Mr. Bedoya. I've already mentioned the need for workplace
privacy law. The other thing we urgently need in my personal
view is some kind of guardrail against automated management.
Increasingly, you're seeing allegations that people are getting
fired automatically on the job because of the way they were
tracked. Maybe it was thought that they were just loafing
around when they were actually going to the bathroom, when they
were actually repairing the machinery on the warehouse floor.
You also see instances where people are increasingly having
their emotions analyzed on the job. This is another call center
example. There was an AI that thought that set off this emotion
distress alarm anytime the caller at the end of the line
started laughing, and so you have technology that doesn't work
as well as people think it does, being used to fire people,
dock their pay, and there need to be guardrails around that
kind of conduct.
Labor unions have often succeeded in putting those
protections in place, but I think everyone needs those kinds of
protections, not just folks who are lucky enough to be in a
union.
Mr. Garcia. Thank you. My time is exhausted.
Mr. Chair, before I turn it back, I seek unanimous consent
to submit an article entitled--
Mr. Fitzgerald. Gentleman is recognized.
Mr. Garcia. Thank you. ``Open AI Peels Back ChatGPT
Safeguards Around Image Creation.''
Mr. Fitzgerald. Without objection.
Mr. Fitzgerald. Gentleman yields back. The gentleman from
Virginia is now recognized for five minutes.
Mr. Cline. Thank you, Mr. Chair.
My goodness, after hearing my friend from California talk,
I'm reminded of the old adage, ``when you're a hammer
everything looks like a nail.'' When you're a proregulatory
Democrat, everything--every industry, every opportunity to
regulate means dive right in and regulate.
Ms. Melugin, can you speak to what would happen to the AI
industry if the U.S. were to adopt heavy-handed regulations.
Which country are we in competition with right now that would
take advantage of that and further run away with development in
AI?
Ms. Melugin. If American innovators are sufficiently
distracted with regulatory hurdles and legal concerns; you are
taking a limited amount of energy and resources away from them
turning out the best product. At that point the alternative for
much of the world will be Chinese products. The Meta question
to keep in mind is do you want the world building on U.S.-based
products or China-based products?
That's not to say that the illegal things that happen in
relation to AI shouldn't be addressed. In large part, due to
already existing law or that Congress might not consider new
privacy legislation, those are all things that can be
considered when problems arise. It's important to not lose
sight of what we're here to do. I don't think that most
Americans are looking for AI to be really, really regulated.
They're looking for AI to be really, really prosperous and
successful.
Mr. Cline. Well, to that end, China has a long history of
intellectual property theft, particularly in AI and
semiconductor technology. How can the U.S. better protect its
AI industry from Chinese espionage, and should we be more
aggressive in sanctioning Chinese companies that benefit from
stolen IP?
Ms. Melugin. Some obviously, but to the extent that all
that can work, I think that's great. At some point, if you're
dealing with bad actors, the way to win is to win. That you
make the biggest mistake by hampering the efforts of U.S.
innovators with all the sort of the side show of all the things
that can possibly go wrong. Of course, there are going to be
challenges and harm that are very real.
People's jobs will be replaced by a certain amount of
automation. I will point out that has been happening since the
dawn of time. It doesn't mean we don't need to have real
attention paid to those circumstances. It's not worth holding
an entire country back in the kind of technology that will
benefit every single American and people around the globe
hopefully from AI products, because we can think of examples of
someone losing their job. It's a tragic situation for that
individual, but I would hope that the overall benefit to the
economy and to society and prosperity of Americans compensates
in some way for that churn. That churn is real, but we can't
use those examples as an excuse to hamper an incredibly
important technology.
Mr. Cline. At the same time, you have big tech companies
arguing that scraping copyrighted material for AI falls under
fair use. This practice could severely undermine artists,
journalists, and content creators should the U.S. continue to
enforce its protections to prevent AI developers from profiting
off stolen intellectual property.
Ms. Melugin. Intellectual property with the forward
movement of technology is a bit of a moving target. It's
certainly an area where Congress can look to revisit and adjust
as necessary. I don't know if we're there yet, but that those
are very valid concerns. I'm certainly not here to argue that
there will be no regulatory or legal reaction to AI. I'm just
here to argue that it's better to go cautiously forward on a
case-by-case basis, and their use of intellectual property
might very well be one of those.
Mr. Cline. Several lawsuits, including The New York Times
v. OpenAI argue that AI models unlawfully use copyrighted
content for training. Given these legal challenges, do you
think courts will ultimately force AI developers to pay
licensing fees, and how might that impact competition and
innovation in the AI industry?
Ms. Melugin. My personal policy is not to comment on what
courts might do. It's possible. I do think one of the pressure
points in AI progress for the U.S. is data sources. I think the
market is innovating even for those synthetic data. There are
other solutions, but new data, more data is an important
balancing act to keep in mind when you're talking about
limiting access to that for maybe the next companies that come
along. I think that there's tradeoffs that need to be weighed
there.
Mr. Cline. There's a balance.
Ms. Melugin. There's a balance, as in most intellectual
property law, I think. Also, I'll quickly make the point that
with data, it's not always more is better. We're finding that
out. Up to a certain point, you need a certain amount, and it
needs to be good and there's engineering progress that can be
made on top of that data, but it's not the end-all/be-all that
AI--it's not the only factor that's important for innovation
and AI.
Mr. Cline. Thank you.
Mr. Fitzgerald. Gentleman yields back.
Mr. Nadler. Mr. Chair, I have a unanimous consent request.
Mr. Fitzgerald. Gentleman is recognized.
Mr. Nadler. Mr. Chair, I ask for unanimous consent to
submit for the record the article by AI Now Institute entitled
``US-China AI Race: AI Policies Industrial Policy,'' that shows
Chinese AI innovation is growing rapidly despite China having
stronger antitrust enforcement than the United States.
Mr. Fitzgerald. Without objection.
Mr. Nadler. Thank you.
Mr. Fitzgerald. The Chair now recognizes the gentleman from
Georgia for five minutes.
Mr. Johnson. Thank you, Mr. Chair.
In an article written by Cecilia Kang of The New York Times
back on last week, she reported that for just over two years,
technology leaders at the forefront of developing artificial
intelligence had made an unusual request of lawmakers. They
wanted Washington to regulate them. The tech executives warned
lawmakers that generative AI, which can produce texts and
images that mimic human creations, had the potential to disrupt
national security, and elections, and could eventually
eliminate millions of jobs. AI could go quite wrong.
Sam Altman, the Chief Executive of OpenAI, testified in
Congress in May 2023; we want to work with the government to
prevent that from happening. For two years, the tech industry--
AI creators have been wanting regulation, but now we have a
group of three lobbyists appear requesting that we not apply
any regulations. What has changed, Ms. Melugin?
Ms. Melugin. Just for recordkeeping purposes, I am not a
lobbyist.
Mr. Johnson. I know you say that you're not, but your
funding--
Ms. Melugin. Well, you have to register to be a lobbyist.
Mr. Johnson. Your funding comes from the industry that
you're here representing today but let me move on. The main
ingredient for artificial intelligence is data, which allows AI
models to learn. Co-President Musk owns an AI company with its
Chatbot Grock, and there is always the problem of AI companies
needing to get a competitive edge to have more data to train
their models, and Co-President Musk is in a unique position
now, because he also runs the so-called Department of
Government Efficiency--DOGE.
In the past 2\1/2\ months, Musk's DOGE team has gained
access to vast data bases at agencies like the Internal Revenue
Service, and the Social Security Administration, which include
our personal information. Just this past weekend, DOGE gained
access to the Federal personnel payroll system.
We know that DOGE has already copied massive amounts of our
personal data onto its servers and using that amount of data to
train his AI model would give Co-President Musk a huge edge
over his competitors.
Now, the administration will tell us that DOGE is being
careful with our personal data, but this is the same
administration whose national security adviser accidentally
invited a journalist to a group chat on a private unsecured
messaging app where the Trump team was discussing war plans and
classified information, and it's the same administration that
conducted national security business over personal Gmail
accounts.
This administration has a culture of playing fast and loose
with secret and private information. Now, how can we trust Elon
Musk with our personal data when he and his AI company have so
much to gain from it financially, Mr. Bedoya?
Mr. Bedoya. I'd love to give a broader answer, if I may. I
do think, speaking just for the FTC, I am very worried about
the information the Commission holds getting in the wrong
hands. The FTC regularly, as a matter of regular course of
business, collects market-sensitive information, information
about competitive strategy, and even top-level information
about certain shops being active at certain times could, for a
person, but definitely for an AI, be used to generate extremely
sensitive inferences about otherwise confidential market
activity.
Mr. Johnson. Elon Musk has his hands around that kind of
information. Commissioner, I talked about the dangers of Co-
President using his personal data, Co-President Musk using our
personal data to train his AI models. Do you also see a threat
to consumers, and what would your concerns be about that?
Mr. Bedoya. The main thing I think about with the use of
data obtained from government agencies being used to train a
model, speaking generally, is let's say a mistake is made and
data is swallowed up into that model that shouldn't be. Twenty
years ago, you go into a system and just delete those files,
but that's not how these models work now. This is like me
saying to you, Congressman, please unlearn my face. Good luck
with that. It doesn't make any sense. So, I'm obviously
oversimplifying a little bit, but it's very difficult to
unscramble those eggs. I'm worried about competitive
information getting into these models, and, therefore, getting
into the wrong hands, even if we realize it after the fact.
That is definitely a real concern.
Mr. Johnson. Those wrong hands are not necessarily in
China. They're right here in DOGE.
Mr. Fitzgerald. Gentleman's time has expired.
Mr. Nadler. Mr. Chair, I have a UC request
Mr. Fitzgerald. Gentleman is recognized.
Mr. Nadler. Thank you, Mr. Chair. I request unanimous
consent to submit a CNBC article entitled, ``Elon Musk who Co-
Founded Firm Behind ChatGPT Warns AI's One of the Biggest Risks
to Civilization,'' where Elon Musk states, quote,
We need to regulate AI safety. It is, I think, actually a
bigger risk to society than cars or planes or medicine.
Mr. Fitzgerald. Without objection, the Chair of the Full
Committee, Mr. Jordan, is now recognized for five minutes.
Chair Jordan. Let me first do unanimous consent in light of
the previous question.
Mr. Fitzgerald. Chair is recognized.
Chair Jordan. I'd like to ask unanimous consent to enter
into the record a story from The Hill, ``The FTC must be held
accountable for its widespread leaks of confidential data.''
Mr. Fitzgerald. Without objection.
Chair Jordan. Mr. Chilson, you worked at the FTC. Is that
right?
Mr. Chilson. That's right.
Chair Jordan. Chief Technologist at the FTC, I think, was
your title.
Mr. Chilson. Acting Chief Technologist, yes.
Chair Jordan. You focused on--I'm looking from the bio
here--the economics in privacy blockchain related issues. Is
that accurate?
Mr. Chilson. That's correct.
Chair Jordan. Worked for the commissioner?
Mr. Chilson. Yes.
Chair Jordan. You're an attorney as well?
Mr. Chilson. I am, yes.
Chair Jordan. You understand the Bill of Rights, the
Constitution, all the things they teach us in law school and
all those important--
Mr. Chilson. I do.
Chair Jordan. Is it appropriate for the FTC to ask an
American company the journalists that they have communicated
with?
Mr. Chilson. I don't think so, sir.
Chair Jordan. I don't think so either, but that's exactly
what happened at the FTC, Mr. Bedoya works at just a couple of
years ago. They sent a letter to Twitter after Mr. Musk had
purchased the business and said it's recently reported that
Twitter has been providing certain third-party journalist
information, and they named the journalists: Bari Weiss, Matt
Tabbi, Michael Shellen-
berger, and Abigal Shrier. They named the journalists, and then
they say this:
Identify all journalists and other members of the media to whom
you've granted any type of access to the companies of
communication.
Is that appropriate?
Mr. Chilson. I don't see what this would have to do with
most of the types of action the FTC would bring.
Chair Jordan. It seems to me this is totally getting at
violating the first amendment and freedom of the press. You've
got a right as a company--people in a company to talk to
journalists. Yet, the FTC wants to know any and every
journalist you've communicated, named four specifically and
wants to know any others you've talked to. What do you think,
Mr. Chilson? Pretty bad stuff that was going on there?
Mr. Chilson. Yep.
Chair Jordan. Let me ask you this, then, too. They also
asked--they wanted to know every reason why Twitter terminated
former Twitter employee Jim Baker. Do you know anything about
that, Mr. Chilson, why they would be asking about that guy?
Mr. Chilson. I don't know much about that, and I don't know
why they would--
Chair Jordan. Well, I got my hunches. Jim Baker was the
General Counsel at the FBI when the FBI was doing all the crazy
things, going to the FISA Court, getting warrants, spying on
the Presidential campaigns. He then goes to work for Twitter.
Mr. Musk buys Twitter and says I'm going to get rid of this
guy, and now the FTC wants to know why. I find that strange
too. Every journalist you've talked to and asked why you got
rid of Jim Baker.
Let me ask you another one. Ernst and Young was named the
assessor in the consent decree focused between the FTC and
Twitter. The assessor, the specific assessor, was a guy, David
Roke. He testified under oath and said, quote,
If Ernst and Young did not include negative findings against
Twitter, the FTC raised the specter that it would create other
challenges for Ernst and Young over time.
Does that seem to be the kind of communication that should be
going from the FTC to the assessor at Ernst and Young, Mr.
Chilson?
Mr. Chilson. Seems like a form of soft coercion.
Chair Jordan. Is that accurate, Mr. Bedoya? Did you guys do
that? Mr. Roke, again, this is under oath in the case, is that
what happened at the FTC? Why would you tell Ernst and Young
you better find out the right thing, you better find something
negative against Twitter or you may not get business with us
over time.
Mr. Bedoya. Thank you, Mr. Chair, for the question. I'm not
familiar with that particular communication, and so given that
I am under oath, I don't want to give a wrong answer here.
Chair Jordan. Are you disputing whether it happened or not?
It's a court filing.
Mr. Bedoya. No, sir, I'm not disputing that. I'm just
saying I'm not familiar with that particular communication, and
like you said, I'm under oath.
Chair Jordan. I'm asking you if it's appropriate or not. It
happened. Here's what he said:
If Ernst and Young did not include negative findings against
Twitter, the FTC raised the specter that it would create other
challenges for Ernst and Young over time.
You hire Ernst and Young to be assessors in these kinds of
consent decree arrangements all the time. You can see the
obvious threat there. I'm just asking if that's appropriate.
Mr. Bedoya. Mr. Chair, who is saying that statement? Who is
writing it?
Chair Jordan. They have the documents that David Roke
communicated in the court that we have.
Mr. Bedoya. Forgive me, Mr. Roke is an Ernst and Young
Executive?
Chair Jordan. Yes. He's the assessor you guys hired.
Mr. Bedoya. He was--I'm sorry, I take this very seriously,
and so what I would ask is if you submit it for the record,
I'll take a good look at it and give you a full and complete
answer, but I'm not familiar with the context of the statement.
Chair Jordan. Why did you have to know about Jim Baker, why
he was terminated? Companies are allowed to fire people.
Mr. Bedoya. Mr. Chair, I did not sign those CIDs, I did not
sign the subpoenas, and so I don't know the exact circumstances
surrounding that individual.
Chair Jordan. OK. OK. No comment there. All right.
Mr. Coniglio, Mr. Bedoya raised three concerns earlier, and
I just have a few seconds. Any of you three can comment on
this. Three warnings that he pointed out. Vertical integration,
AI keeping private data private, and then the surveillance
concern. Do you have a different perspective on those three
concerns? Or tell me what your thoughts on the three warnings
he laid out in his testimony?
Mr. Coniglio. Thank you for the question. I can start with
the first one about AI partnerships and vertical integration
and any competitive conduct. I think that the concerns about
foreclosure, which are the typical anticompetitive harm that
you think about when you go to vertical integration are
certainly very premature at this stage. We see robust
competition at the foundation model level and the cloud level,
which suggests that a vertical integration between, let's say,
a cloud company and a foundation model is unlikely to result in
any sort of exclusionary conduct.
With respect to the partnerships and the horizontal, so to
speak, reduction of competition between let's say a big tech
firm that may have a partnership and the idea of being that
because they are investing in these companies, they're less
likely to compete, again, it's unlikely that's going to result
in a substantial lessening of competition. Again, because of
the robust competition that exists at the foundation model
level, but also because of the procompetitive benefits that
these partnerships can have, specifically providing internet--
excuse me, AI foundation model startups with the sort of
patient capital they need to grow sustainably and actually
innovate faster.
Chair Jordan. OK. I know I'm out of time, but can you talk
about the privacy issue really quick, too? Anyone of you want
to talk on privacy? I have concerns about this whole facial
recognition stuff and some of the examples that Mr. Bedoya gave
in his testimony, so give me your quick thoughts on that, and
then I'll be happy to yield back, Mr. Chair.
Mr. Chilson. In many ways, the privacy issue and artificial
intelligence, especially these deep learning models, is much
more complicated than it is in the form of social media, and
so, when you train these models, what you get is something that
does not have any of the actual training data in it. You can
never look into a model and see the specific training that was
in there.
The privacy implications here are once more on the
inference side and not about actual data being in these, and so
it's much more like remembering that something happened, but
not remembering it precisely, and therefore, I think--
Mr. Fitzgerald. Gentleman's--
Mr. Chilson. Sorry. I think the privacy concerns are less.
Mr. Fitzgerald. Gentleman's time has expired.
Chair Jordan. Thank you, Mr. Chair.
Mr. Fitzgerald. Gentleman yields back. Ask unanimous
consent to enter into the record the letter from Gary Shapiro
and Kinsey Fabrizio of CTA dated April 2, 2025. Also, the
statement of Jennifer Huddleston from Cato dated April 2, 2025,
and the letter from Graham Dufault of the App Association dated
April 2, 2025. Without objection.
Mr. Garcia. Mr. Chair.
Mr. Fitzgerald. Gentleman from Illinois has a UC request.
Mr. Garcia. Yes. Thank you. Chair, I request unanimous
consent to add to the record a Tech Policy Press article
entitled, ``DOGE Plan to Push AI Across the U.S. Federal
Government is Wildly Dangerous. ``
Mr. Fitzgerald. Without objection.
Mr. Garcia. Thank you.
Mr. Fitzgerald. The Ranking Member of the Full Committee is
now recognized for five minutes.
Mr. Raskin. Thank you kindly, Mr. Chair, and thanks to our
witnesses today.
It's hard to think of an antitrust matter more pressing
than concentration in the AI market. The dominance of a handful
of AI companies presents a long-term problem for our national
security, our democracy, our privacy, and basic fairness in the
marketplace. President Trump has plunged our government into
chaos and disabled the Federal agencies that we depend on to
address such problems. It will be difficult to effectively
respond to dangerous market concentration in the assembly of
all human knowledge if we have governmental power that is
dangerously concentrated itself in the President to the
exclusion of Congress as the law-making branch and the
Judiciary as the protector of constitutional rights and values.
Over a century ago, Congress established the Federal Trade
Commission. It was created as an independent agency to enforce
antitrust and consumer protection laws outside of the
controlling pressure and power agendas of any single President
or party. For more than a century, the FTC has protected
Americans against scams, fraud, and other unfair and deceptive
practices. It also promoted and protected competition in the
marketplace, which benefits consumers, workers, small
businesses driving innovation, and business without monopoly.
Congress designed the FTC to be led by five commissioners
across party lines who served seven-year terms are subject to
Senate confirmation and can only be fired for cause. Congress
structured this independence to ensure that the FTC would serve
the nonpartisan interest of the American people and be
insulated from partisan political control.
Over 90 years ago, the independence of these commissioners
was tested and the Supreme Court unanimously reaffirmed that
the President cannot unilaterally fire an independent FTC
commissioner without cause. Unitary Executive power violated
the careful Federal design and infringed Congress's lawmaking
power to establish the FTC as an effective and balanced
instrument for the public good. Yet, President Trump chose to
violate Federal law and defy the Court's longstanding precedent
by attempting to fire without cause the two Democratic members
of the FTC. He did so to end the nonpartisan independence of
the FTC, and to make sure it responds to his momentary
political demands above the national public interest as
contemplated by Congress.
Accordingly, the commissioners have brought suit to reverse
their unlawful firing, and two, as Commissioner Slaughter
stated,
Make sure that the work of the Commission is done without fear
or favor and more specifically that we can take on the biggest
companies in America without fear of getting fired for failure
to do a favor to the President's friends who are donors or
corporate allies.
The strongest Federal rules on AI were established by
President Biden in a 2023 EO. The Trump Administration reversed
that Executive Order, which laid down the guardrails for the AI
industry. Not content with this rollback, the President has now
hobbled and undermined the lead cop on the AI beat, the FTC,
which has for years been at the forefront of holding big tech,
including AI, to account.
It's hard to debate the proper scope of regulation in this
political space when Donald Trump has reportedly granted Elon
Musk unlimited access to our most private and sensitive
records, data that's undoubtedly invaluable to X AI and his AI
company. Today, none of us here know whether Musk is using our
data, classified information, and other sensitive data from
data bases across government to train his company's AI models
in gaining an unfair advantage over competitors and against
citizens. Musk has refused to answer many requests on this
topic, and now he's crippled the agency charged with
investigating these very tech abuses.
Even if we came together and found agreement on regulatory
reforms and safeguards for AI development, we still now lack an
expert independent agency to enforce our decisions. The
President and Musk are ushering a new regime in which
monarchical power, cronyism, corruption, and sycophancy rule.
He's using his office to enrich himself, his friends, and
his donors at the expense of the people. The attempted firing
of two FTC commissioners and his subsequent removal of any FTC
resource that might serve as a check on big tech illustrate how
the President is dismantling the machinery of government to
serve his own end.
Our Constitutional order is under attack to concentrate all
political power in the President. It's the political equivalent
of what we're trying to prevent in the economic sphere. I hope
that our witnesses can give us their best insight as to how to
deal with this problem, but we're facing a structural crisis
now in our government.
Thank you, Mr. Chair. I yield back.
Mr. Fitzgerald. Gentleman yields back.
Chair Jordan. Mr. Chair.
Mr. Fitzgerald. The Chair is recognized for a UC request.
Mr. Jordan. I'd ask unanimous consent to enter into the
record the letter from August 26th of last year sent by one of
those oligarchs that the Ranking Member just put up there, Mr.
Zuckerberg, the letter he sent to the Committee saying that
they, in fact, censored Americans at the request of the Biden
Administration. I yield back.
Mr. Fitzgerald. We thank the witnesses for appearing before
the Committee today. Without objection, all Members will have
five legislative days to submit additional written questions
for the witnesses or additional materials for the record.
Without objection, this hearing is adjourned.
[Whereupon, at 11:49 a.m., the Subcommittee was adjourned.]
All materials submitted for the record by Members of the
Subcommittee on the Administrative State, Regulatory Reform,
and Antitrust can be found at: https://docs.house.gov/
Committee/
Calendar/ByEvent.aspx?EventID=118081.
[all]