[House Hearing, 116 Congress]
[From the U.S. Government Publishing Office]
AMERICANS AT RISK: MANIPULATION AND
DECEPTION IN THE DIGITAL AGE
=======================================================================
HEARING
BEFORE THE
SUBCOMMITTEE ON CONSUMER PROTECTION AND COMMERCE
OF THE
COMMITTEE ON ENERGY AND COMMERCE
HOUSE OF REPRESENTATIVES
ONE HUNDRED SIXTEENTH CONGRESS
SECOND SESSION
__________
JANUARY 8, 2020
__________
Serial No. 116-86
[GRAPHIC NOT AVAILABLE IN TIFF FORMAT]
Printed for the use of the Committee on Energy and Commerce
govinfo.gov/committee/house-energy
energycommerce.house.gov
__________
U.S. GOVERNMENT PUBLISHING OFFICE
44-716 PDF WASHINGTON : 2021
-----------------------------------------------------------------------------------
COMMITTEE ON ENERGY AND COMMERCE
FRANK PALLONE, Jr., New Jersey
Chairman
BOBBY L. RUSH, Illinois GREG WALDEN, Oregon
ANNA G. ESHOO, California Ranking Member
ELIOT L. ENGEL, New York FRED UPTON, Michigan
DIANA DeGETTE, Colorado JOHN SHIMKUS, Illinois
MIKE DOYLE, Pennsylvania MICHAEL C. BURGESS, Texas
JAN SCHAKOWSKY, Illinois STEVE SCALISE, Louisiana
G. K. BUTTERFIELD, North Carolina ROBERT E. LATTA, Ohio
DORIS O. MATSUI, California CATHY McMORRIS RODGERS, Washington
KATHY CASTOR, Florida BRETT GUTHRIE, Kentucky
JOHN P. SARBANES, Maryland PETE OLSON, Texas
JERRY McNERNEY, California DAVID B. McKINLEY, West Virginia
PETER WELCH, Vermont ADAM KINZINGER, Illinois
BEN RAY LUJAN, New Mexico H. MORGAN GRIFFITH, Virginia
PAUL TONKO, New York GUS M. BILIRAKIS, Florida
YVETTE D. CLARKE, New York, Vice BILL JOHNSON, Ohio
Chair BILLY LONG, Missouri
DAVID LOEBSACK, Iowa LARRY BUCSHON, Indiana
KURT SCHRADER, Oregon BILL FLORES, Texas
JOSEPH P. KENNEDY III, SUSAN W. BROOKS, Indiana
Massachusetts MARKWAYNE MULLIN, Oklahoma
TONY CARDENAS, California RICHARD HUDSON, North Carolina
RAUL RUIZ, California TIM WALBERG, Michigan
SCOTT H. PETERS, California EARL L. ``BUDDY'' CARTER, Georgia
DEBBIE DINGELL, Michigan JEFF DUNCAN, South Carolina
MARC A. VEASEY, Texas GREG GIANFORTE, Montana
ANN M. KUSTER, New Hampshire
ROBIN L. KELLY, Illinois
NANETTE DIAZ BARRAGAN, California
A. DONALD McEACHIN, Virginia
LISA BLUNT ROCHESTER, Delaware
DARREN SOTO, Florida
TOM O'HALLERAN, Arizona
------
Professional Staff
JEFFREY C. CARROLL, Staff Director
TIFFANY GUARASCIO, Deputy Staff Director
MIKE BLOOMQUIST, Minority Staff Director
Subcommittee on Consumer Protection and Commerce
JAN SCHAKOWSKY, Illinois
Chairwoman
KATHY CASTOR, Florida CATHY McMORRIS RODGERS, Washington
MARC A. VEASEY, Texas Ranking Member
ROBIN L. KELLY, Illinois FRED UPTON, Michigan
TOM O'HALLERAN, Arizona MICHAEL C. BURGESS, Texas
BEN RAY LUJAN, New Mexico ROBERT E. LATTA, Ohio
TONY CARDENAS, California, Vice BRETT GUTHRIE, Kentucky
Chair LARRY BUCSHON, Indiana
LISA BLUNT ROCHESTER, Delaware RICHARD HUDSON, North Carolina
DARREN SOTO, Florida EARL L. ``BUDDY'' CARTER, Georgia
BOBBY L. RUSH, Illinois GREG GIANFORTE, Montana
DORIS O. MATSUI, California GREG WALDEN, Oregon (ex officio)
JERRY McNERNEY, California
DEBBIE DINGELL, Michigan
FRANK PALLONE, Jr., New Jersey (ex
officio)
C O N T E N T S
----------
Page
Hon. Jan Schakowsky, a Representative in Congress from the State
of Illinois, opening statement................................. 2
Prepared statement........................................... 2
Hon. Cathy McMorris Rodgers, a Representative in Congress from
the State of Washington, opening statement..................... 3
Prepared statement........................................... 5
Hon. Frank Pallone, Jr., a Representative in Congress from the
State of New Jersey, opening statement......................... 6
Prepared statement........................................... 7
Hon. Greg Walden, a Representative in Congress from the State of
Oregon, opening statement...................................... 8
Prepared statement........................................... 10
Witnesses
Monika Bickert, Vice President of Global Policy Management,
Facebook....................................................... 12
Prepared statement........................................... 14
Answers to submitted questions............................... 108
Joan Donovan, Ph.D., Director, Technology and Social Change
Project, Shorenstein Center on Media, Politics and Public
Policy, Harvard Kennedy School................................. 19
Prepared statement........................................... 21
Answers to submitted questions \1\........................... 124
Justin (Gus) Hurwitz, Director of Law and Economics Programs,
International Center for Law and Economics..................... 26
Prepared statement........................................... 29
Answers to submitted questions............................... 129
Tristan Harris, President and Cofounder, Center for Humane
Technology..................................................... 50
Prepared statement........................................... 52
Answers to submitted questions............................... 137
Submitted Material
Letter of January 8, 2020, from Kerri Wood Einertson, National
Director, Government Affairs and Public Policy, SAG-AFTRA, to
Ms. Shakowsky and Mrs. Rodgers, submitted by Ms. Schakowsky.... 104
Letter of January 8, 2020, from Jeff Westling, Technology and
Innovation Policy Fellow, R Street Institute, to Ms. Shakowsky
and Mrs. Rodgers, submitted by Ms. Schakowsky.................. 106
Report of June 2019, ``Are Deep Fakes a Shallow Concern? A
Critical Analysis of the Likely Societal Reactions to Deep
Fakes,'' submitted by Ms. Schakowsky \2\
Report, ``Facebook's Black Market in Antiquities,'' by Amr Al-Azm
and Katie A. Paul, submitted by Ms. Schakowsky \3\
----------
\1\ Dr. Donovan did not answer submitted questions for the record by
the time of publication.
\2\ The report has been retained in committee files and also is
available at https://docs.house.gov/meetings/IF/IF17/20200108/110351/
HHRG-116-IF17-20200108-SD005.pdf.
\3\ The report has been retained in committee files and also is
available at https://docs.house.gov/meetings/IF/IF17/20200108/110351/
HHRG-116-IF17-20200108-SD006.pdf.
AMERICANS AT RISK: MANIPULATION AND DECEPTION IN THE DIGITAL AGE
----------
WEDNESDAY, JANUARY 8, 2020
House of Representatives,
Subcommittee on Consumer Protection and Commerce,
Committee on Energy and Commerce,
Washington, DC.
The subcommittee met, pursuant to call, at 10:32 a.m., in
the John D. Dingell Room 2123, Rayburn House Office Building,
Hon. Jan Schakowsky (chairwoman of the subcommittee) presiding.
Members present: Representatives Schakowsky, Castor,
Veasey, Kelly, O'Halleran, Lujan, Cardenas, Blunt Rochester,
Soto, Matsui, McNerney, Dingell, Pallone (ex officio), Rodgers
(subcommittee ranking member), Burgess, Latta, Guthrie,
Bucshon, Hudson, Carter, and Walden (ex officio).
Also present: Representative Clarke.
Staff present: Jeffrey C. Carroll, Staff Director; Evan
Gilbert, Deputy Press Secretary; Lisa Goldman, Senior Counsel;
Waverly Gordon, Deputy Chief Counsel; Tiffany Guarascio, Deputy
Staff Director; Alex Hoehn-Saric, Chief Counsel, Communications
and Consumer Protection; Zach Kahan, Outreach and Member
Service Coordinator; Joe Orlando, Executive Assistant; Alivia
Roberts, Press Assistant; Chloe Rodriguez, Policy Analyst;
Sydney Terry, Policy Coordinator; Rebecca Tomilchik, Staff
Assistant; Anna Yu Professional Staff Member; Mike Bloomquist,
Minority Staff Director; S.K. Bowen, Minority Press Assistant;
William Clutterbuck, Minority Staff Assistant; Jordan Davis,
Minority Senior Advisor; Tyler Greenberg, Minority Staff
Assistant; Peter Kielty, Minority General Counsel; Ryan Long,
Minority Deputy Staff Director; Mary Martin, Minority Chief
Counsel, Energy, and Environment and Climate Change; Brandon
Mooney, Minority Deputy Chief Counsel, Energy; Brannon Rains,
Minority Legislative Clerk; Zack Roday, Minority Director of
Communications; and Peter Spencer, Minority Senior Professional
Staff Member, Environment and Climate Change.
Ms. Schakowsky. Good morning, everyone. The Subcommittee on
Consumer Protection and Commerce will now come to order. We
will begin with Member statements, and I will begin by
recognizing myself for 5 minutes.
OPENING STATEMENT OF HON. JAN SCHAKOWSKY, A REPRESENTATIVE IN
CONGRESS FROM THE STATE OF ILLINOIS
Good morning, and thank you for joining us here today.
Given what is going on in the world, it is really impressive to
see the turnout that is here today, and I welcome everyone.
In the two-plus decades since the creation of the internet,
we have seen life for Americans and their families transformed
in many positive ways. The internet provides new opportunities
for commerce, education, information, and connecting people.
However, along with these many new opportunities, we have
seen new challenges as well. Bad actors are stocking the online
marketplace, using deceptive techniques to influence consumers,
deceptive designs to fool them into giving away personal
information, stealing their money, and engaging in other unfair
practices.
The Federal Trade Commission works to protect Americans
from many unfair and deceptive practices, but a lack of
resources, authority, and even a lack of will has left many
American consumers feeling helpless in this digital world.
Adding to that feeling of helplessness, new technologies are
increasing the scope and scale of the problem. Deepfakes,
manipulation of video, dark patterns, bots, and other
technologies are hurting us in direct and indirect ways.
Congress has, unfortunately, taken a laissez faire approach
to regulation of unfair and deceptive practices online over the
past decade, and platforms have let them flourish. The result
is Big Tech failed to respond to the grave threats posed by
deepfakes, as evidenced by Facebook scrambling to announce a
new policy that strikes me as wholly inadequate--we will talk
about that later--since it would have done nothing to prevent
the video of Speaker Pelosi that amassed millions of views and
prompted no action by the online platform. Hopefully, our
discussion today can change my mind about that.
Underlying all of this is Section 230 of the Communications
Decency Act, which provides online platform links like Facebook
a legal liability shield for third-party content. Many have
argued that this liability shield results in online platforms
not adequately policing their platforms, including online
piracy and extremist content.
Thus, here we are, with Big Tech wholly unprepared to
tackle the challenges we face today. A top-line concern for
this subcommittee must be to protect consumers, regardless of
whether they are online or not. For too long, Big Tech has
argued that e-commerce and digital platforms deserve special
treatment and a light regulatory touch.
We are finding out that consumers can be harmed as easily
online as in the physical world, and in some cases that online
dangers are greater. It is incumbent on us in this subcommittee
to make clear that the protections that apply to in-person
commerce also apply to virtual space.
[The prepared statement of Ms. Schakowsky follows:]
Prepared Statement of Hon. Jan Schakowsky
Good morning and thank you for joining us here today. In
the two plus decades since the creation of the internet, we
have seen life for Americans and their families transformed in
many positive ways. The internet provides new opportunities for
commerce, education, information, and connecting people.
However, along with these many new opportunities, we have
seen new challenges. Bad actors are stalking the online
marketplace using deceptive techniques to influence consumers,
deceptive designs to fool them into giving away personal
information, stealing their money, and engaging in other unfair
practices.
The Federal Trade Commission works to protect Americans
from many unfair and deceptive practices, but a lack of
resources, authority, and even a lack of will has left many
American consumers feeling helpless in the digital world.
Adding to that feeling of helplessness, new technologies
are increasing the scope and scale of the problem. Deepfakes,
manipulated video, dark patterns, bots, and other technologies
are hurting us in direct and indirect ways.
Congress has unfortunately taken a laissez faire approach
to regulating unfair and deceptive practices online over the
past decade and platforms have let them flourish.
The result is big tech failed to respond to the grave
threat posed by deep-fakes, as evidenced by Facebook scrambling
to announce a new policy that strikes me as wholly inadequate,
since it would have done nothing to prevent the altered video
of Speaker Pelosi that amassed millions of views and prompted
no action by the online platform. Hopefully our discussion
today can change my mind.
Underlying all of this is Section 230 of the Communications
Decency Act, which provided online platforms like Facebook a
legal liability shield for 3rd party content. Many have argued
that this liability shield resulted in online platforms not
adequately policing their platforms, including online piracy
and extremist content. Thus, here we are, with Big Tech wholly
unprepared to tackle the challenges we face today.
A topline concern for this subcommittee must be to protect
consumers regardless of whether they are online or not. For too
long, Big Tech has argued that e-commerce and digital platforms
deserved special treatment and a light regulatory touch. We are
finding out that consumers can be harmed as easily online as in
the physical world. And in some cases, the online dangers are
greater. It's incumbent on this subcommittee to make clear that
protections that apply to in-person commerce also apply in the
virtual space. I thank the witnesses for their testimony, and I
recognize Ranking Member Rodgers for 5 minutes.
Ms. Schakowsky. I thank the witnesses for their testimony
today, and I recognize Ranking Member Rodgers for 5 minutes.
OPENING STATEMENT OF HON. CATHY McMORRIS RODGERS, A
REPRESENTATIVE IN CONGRESS FROM THE STATE OF WASHINGTON
Mrs. Rodgers. Thank you. Thank you, Chair Schakowsky. Happy
New Year, everyone. Welcome to our witnesses. I appreciate the
chair leading this effort today to highlight online deception.
I do want to note that last Congress, Chairman Walden also
held several hearings on platform responsibility.
Disinformation is not a new problem. It was also an issue 130
years ago when Joseph Pulitzer and the New York World and
William Randolph Hearst and The New York Journal led the age
of, quote, ``yellow journalism.'' Just like clickbait on online
platforms today, fake and sensational headlines sold newspapers
and boosted advertising revenue. With far more limited sources
of information available in the 1890s, the American people lost
trust in the media. To rebuild trust, newspapers had to clean
up their act. Now the Pulitzer is associated with something
very different.
I believe we are at a similar inflection point today. We
are losing faith in sources we can trust online. To rebuild it,
this subcommittee, our witness panel and members of the media
are putting the spotlight on abuses and deception.
Our committee's past leadership and constructive debates
have already led to efforts by platforms to take action. Just
this week, Facebook announced a new policy to combat deepfakes,
in part, by utilizing artificial intelligence. I appreciate Ms.
Bickert for being here to discuss this in greater detail.
Deepfakes and disinformation can be handled with innovation and
empowering people with more information.
On the platforms they choose and trust, it makes far more
productive outcomes when people can make the best decisions for
themselves, rather than relying on the government to make
decisions for them. That is why we should be focusing on
innovation for major breakthroughs, not more regulations or
government mandates.
As we discuss ways to combat manipulation online, we must
ensure that America will remain the global leader in AI
development. There is no better place in the world to raise
people's standard of living and make sure that this technology
is used responsibly.
Software is already available to face swap, lip sync, and
create facial reenactment to fabricate content. As frightening
as it is, we can also be using AI to go after the bad actors
and fight fire with fire. We cannot afford to shy away from it,
because who would you rather lead the world in machine learning
technology: America or China? China is sharing its AI
surveillance technology with other authoritarian governments,
like Venezuela. It is also using its technology to control and
suppress ethnic minorities, including the Uighurs in Chinese
concentration camps.
The New York Times has reported just last month that China
is collecting DNA samples and could be using this data to
create images of faces. Could China be building a tool to
further track and crack down on minorities and political
dissidents? Imagine the propaganda and lies it could develop
with this technology behind the Great Chinese Firewall, where
there is no free speech or an independent press to hold the
Communist Party accountable.
That is why America must lead the world in AI development.
By upholding our American values, we can use this as a force
for good and save people's lives. For example, AI technology
and deep learning algorithms can help us detect cancers earlier
and more quickly. Clinical trials are already underway and
making major breakthroughs to diagnose cancers.
The continued leadership of our innovators is crucial to
make sure that we have the tools to combat online deception. To
win the future in a global economy, America should be writing
the rules for this technology so that real people, not an
authoritarian state like China, are empowered.
I am also glad that we are putting a spotlight on dark
patterns. Deceptive laws, fake reviews, and bots are the latest
version of robocall scams. I am pleased that the FTC has used
its Section 5 authority to target this fraud and protect
people. We should get their input as to how we discuss how to
handle dark patterns.
We also must be careful where we legislate so that we don't
harm the practices that people enjoy. A heavy-handed regulation
will make it impossible for online retailers to provide
discounts. This would especially hurt lower- and middle-income
families. In a digital marketplace, services people enjoy
should not get swallowed up by strict definition of a dark
pattern. How we make these distinctions is important, so I look
forward to today's discussion.
I want to thank the panel, and I yield back.
[The prepared statement of Mrs. Rodgers follows:]
Prepared Statement of Hon. Cathy McMorris Rodgers
Thank you, Chair Schakowsky and welcome to our witnesses.
I appreciate your work to highlight online deception.
Last Congress, Chairman Walden led several hearings on
platform responsibility, before it became the popular cause it
is today.
Disinformation is not a new problem.
It was also an issue 130 years ago when Joseph Pulitzer and
the New York World and William Randolph Hearst and the New York
Journal led the age of quote ``yellow journalism.''
Just like ``clickbait'' on online platforms today, fake and
sensational headlines sold newspapers and boosted advertising
revenue.
With far more limited sources of information available in
the 1890s, the American public lost trust in the media.
To rebuild trust, newspapers had to clean up their act.
Now the name Pulitzer is associated with something very
different.
I believe we are at a similar inflection point today.
We are losing faith in sources we can trust online.
To rebuild it.this subcommittee, our witness panel, and
members of the media are putting the spotlight on abuses and
deception.
Our committee's past leadership and constructive debates
have already led to efforts by platforms to take action.
Just this week Facebook announced a new policy to combat
deepfakes, in part by utilizing artificial intelligence.
I appreciate Ms. Bickert for coming here to discuss this in
greater detail.
``Deepfakes'' and disinformation can be handled with
innovation and empowering people with MORE information.
On the platforms they choose and trust, it's a far more
productive outcome when people can make the best decisions for
themselves rather than relying on the government to make
decisions for them.
That's why we should be focusing on innovation for major
breakthroughs. Not more regulations or government mandates.
As we discuss ways to combat manipulation online, we must
ensure America will remain the global leader in AI development.
There's no better place in the world to raise people's
standard of living and make sure this technology is used
responsibly.
Software is already available to face swap, lip sync, and
create facial reenactment to fabricate content.
As frightening as this is, we can also be using AI to go
after bad actors and fight fire with fire.
We cannot afford to shy away from it because who would you
rather lead the world in machine learning technology?
America or China?
China is sharing its AI-surveillance technology with other
authoritarian governments like in Venezuela
It's also using this technology to control and suppress
ethnic minorities, including the Uighurs in Chinese
concentration camps.
The New York Times reported just last month that China is
collecting DNA samples of Uighurs and could be using this data
to create images of their faces.
Could China be building a tool to further track and crack
down on minorities and political dissidents?
Imagine the propaganda and lies they could develop with
this technology behind the Great Chinese Firewall, where
there's no free speech or an independent press to hold the
Communist Party accountable.
This is why America must lead the world in AI development.
By upholding our American values, we can use this as a
force for good and save people's lives.
For example, AI technology and deep-learning algorithms can
help us detect cancers earlier and more quickly.
Clinical trials are already making major breakthroughs to
diagnose cancers.
The continued leadership of our innovators is crucial to
make sure we have tools to combat online deception too.
I applaud the Trump administration for their forward-
thinking leadership in setting a light-touch framework for
encouraging continued, responsible American innovation in AI.
To win the future in a global economy, America should be
writing the rules for this technology so real people--not an
authoritarian state like China--are empowered.
I'm also glad we're putting a spotlight on ``dark
patterns.''
Deceptive ads, fake reviews, and bots are the latest
version of robocall scams.
I'm pleased that the FTC has used its Section 5 authority
to target this fraud and protect people.
We should get their input as we discuss how to handle dark
patterns.
We must be careful where we legislate so we don't harm
practices that people enjoy.
A heavy-handed regulation will make it impossible for
online retailers to provide discounts.
This would especially hurt lower- and middle-income
families.
In the digital marketplace, services people enjoy should
not get swallowed up by a strict definition of a ``dark
pattern''.
How we make these distinctions is important.
I'm looking forward to today's discussion. Thank you again
to our panel. Thank you, and I yield back.
Ms. Schakowsky. The gentlelady yields back.
And the Chair now recognizes Mr. Pallone, chair of the full
committee, for 5 minutes for his opening statement.
OPENING STATEMENT OF HON. FRANK PALLONE, Jr., A REPRESENTATIVE
IN CONGRESS FROM THE STATE OF NEW JERSEY
Mr. Pallone. Thank you, Madam Chair.
Americans increasingly rely on the internet for fundamental
aspects of their daily lives. Consumers shop online for
products ranging from groceries to refrigerators. They use the
internet to telecommute or to check the weather and traffic
before leaving for the office, and they use social media
networks to connect with family and friends, and as a major
source of news and information.
When consumers go online, they understandably assume that
the reviews of the products that they buy are real, that the
people on the social networks are human, and that the news and
information they are reading is accurate. But, unfortunately,
that is not always the case. Online actors, including nation-
states, companies, and individual fraudsters, are using online
tools to manipulate and deceive Americans. While some methods
of deception are well-known, many are new and sophisticated,
fooling even the most savvy consumers.
Today, technology has made it difficult, if not impossible,
for typical consumers to recognize what is real from what is
fake. And why exactly are people putting so much effort into
the development and misuse of technology? Because they know
that trust is the key to influencing and taking advantage of
people, whether for social, monetary, or political gain. If bad
actors can make people believe a lie, then they can manipulate
us into taking actions we wouldn't otherwise take.
In some instances, we can no longer even trust our eyes.
Videos can be slowed to make someone appear intoxicated. Faces
can be Photoshopped onto someone else's body. Audio can be
edited in a way that a person's words are basically taken out
of context. And the extent of such manipulation has become
extreme. Machine-learning algorithms can now create completely
fake videos, known as deepfakes, that look real. Deepfakes can
show real people saying or doing things that they never said or
did.
For example, face-swapping technology has been used to
place actor Nicolas Cage into movies where he never was. Actor/
director Jordan Peele created a deepfake supposedly showing
President Obama insulting President Trump.
The most common use of deepfakes is nonconsensual
pornography, which has been used to make it appear as if
celebrities have been videotaped in compromising positions. And
deepfake technology was also used to humiliate a journalist
from India who was reporting on an 8-year-old rape victim.
Advances in algorithms are also behind the glut of social
media bots, automated systems that interact on social media as
if they were real people. These bots are used by companies and
other entities to build popularity of brands and respond to
consumer service requests. Even more alarming is the use of
these bots by both state and nonstate actors to spread
disinformation, which can influence the very fabric of our
society and our politics.
And manipulation can be very subtle. Deceptive designs,
sometimes called dark patterns, capitalize on knowledge of our
senses, operate to trick us into making choices that benefit
the business. Have you ever tried to unsubscribe from a mailing
list and there is a button to stay subscribed that is bigger
and more colorful than the unsubscribe button? And that is
deceptive design. Banner ads have been designed with black
spots that look like dirt or hair on the screen to trick you
into tapping the ``add'' on your smartphone. And there are so
many other examples.
And since these techniques are designed to go unnoticed,
most consumers have no idea they are happening. In fact, they
are almost impossible for experts in types of techniques to
detect. And, while computer scientists are working on
technology that can help detect each of these deceptive
techniques, we are in a technological arms race. As detection
technology improves, so does the deceptive technology.
Regulators and platforms trying to combat deception are left
playing Whac-a-mole.
Unrelenting advances in these technologies and their abuse
raise significant questions for all of us. What is the
prevalence of these deceptive techniques? How are these
techniques actually affecting our actions and decisions? What
steps are companies and regulators taking to mitigate consumer
fraud and misinformation?
So I look forward to beginning to answer these questions
with our expert witness panel today so we can start to provide
more transparency and tools for consumers to fight
misinformation and deceptive practices.
And, Madam Chair, I just want to say I think this is a very
important hearing. I was just telling my colleague, Kathy
Castor, this morning about a discussion that we had at our
chairs meeting this morning, where the topic was brought up.
And I said, ``Oh, you know, we are having a hearing on this
today.'' So this is something a lot of Members and, obviously,
the public care about. So thank you for having the hearing
today.
I yield back.
[The prepared statement of Mr. Pallone follows:]
Prepared Statement of Hon. Frank Pallone, Jr.
Americans increasingly rely on the internet for fundamental
aspects of their daily lives. Consumers shop online for
products ranging from groceries to refrigerators. They use the
internet to telecommute or to check the weather and traffic
before leaving for the office. And they use social media
networks to connect with family and friends and as a major
source of news and information.
When consumers go online they understandably assume that
the reviews of the products they buy are real, that the people
on their social networks are human, and that the news and
information they are reading is accurate. Unfortunately, that
is not always the case.
Online actors, including nation-states, companies, and
individual fraudsters, are using online tools to manipulate and
deceive Americans. While some methods of deception are well
known, many are new and sophisticated, fooling even the most
savvy consumers. Today, technology has made it difficult, if
not impossible, for typical consumers to recognize what's real
from what's fake.
And why exactly are people putting so much effort into the
development and misuse of this technology? Because they know
that trust is the key to influencing and taking advantage of
people. Whether for social, monetary, or political gain, if bad
actors can make people believe a lie, they can manipulate us
into taking actions we wouldn't otherwise take.
In some instances, we can no longer even trust our eyes.
Videos can be slowed to make someone appear intoxicated. Faces
can be Photoshopped onto someone else's body. Audio can be
edited in a way that takes a person's words out of context.
The extent of such manipulation has become extreme. Machine
learning algorithms can now create completely fake videos,
known as deepfakes, that look real. Deepfakes can show real
people saying or doing things they never said or did.
For example, face-swapping technology has been used to
place actor Nicolas Cage into movies he was never in. Actor-
director Jordan Peele created a deepfake supposedly showing
President Obama insulting President Trump. The most common use
of deepfakes is nonconsensual pornography, which has been used
to make it appear as if celebrities have been videotaped in
compromising positions. Deepfake technology was also used to
humiliate a journalist from India who was reporting on an 8-
year-old rape victim.
Advances in algorithms are also behind the glut of social
media bots, automated systems that interact on social media as
if they were real people. These bots are used by companies and
other entities to build popularity of brands and respond to
customer service requests. Even more alarming is the use of
these bots by both state and nonstate actors to spread
disinformation, which can influence the very fabric of our
societies and our politics.
And manipulation can be very subtle. Deceptive design,
sometimes called ``dark patterns,'' capitalize on knowledge of
how our senses operate to trick us into making choices that
benefit the business. Have you ever tried to unsubscribe from a
mailing list and there's a button to stay subscribed that's
bigger and more colorful than the unsubscribe button? That's
deceptive design. Banner ads have been designed with black
spots that look like dirt or a hair on the screen to trick you
into tapping the ad on your smartphone. And there are many more
examples.
Since these techniques are designed to go unnoticed, most
consumers have no idea they are happening. In fact, they are
almost impossible for experts in types of techniques to detect.
While computer scientists are working on technology that
can help detect each of these deceptive techniques, we are in a
technological arms race. As detection technology improves, so
does the deceptive technology. Regulators and platforms trying
to combat deception are left playing Whac-a-Mole.
Unrelenting advances in these technologies and their abuse
raise significant questions for all of us. What is the
prevalence of these deceptive techniques? How are these
techniques actually affecting our actions and decisions? What
steps are companies and regulators taking to mitigate consumer
fraud and misinformation?
I look forward to beginning to answer these questions with
our expert witness panel today so that we can start to provide
more transparency and tools for consumers to fight
misinformation and deceptive practices.
Ms. Schakowsky. The gentleman yields back.
And now the Chair recognizes Mr. Walden, the ranking member
of the full committee, for 5 minutes for his opening statement.
OPENING STATEMENT OF HON. GREG WALDEN, A REPRESENTATIVE IN
CONGRESS FROM THE STATE OF OREGON
Mr. Walden. Good morning, Madam Chair. Thanks for having
this hearing and welcome everyone in. I guess this is the
second hearing of the new year. There is one that started
earlier upstairs, but we welcome you all to hear this important
topic and glad to hear from our witnesses today, even those who
I am told have health issues this morning, but thanks for being
here.
As with anything, the internet presents bad actors with
those seeking to harm others some ample opportunities to
manipulate the users and take advantage of consumers, which
often tend to be some of the most vulnerable in the population.
Arguably, the digital ecosystem is such that harmful acts are
easily exacerbated, and as we all know, false information or
fake videos spread at breakneck speeds.
That is why, when I was chairman of this committee, we
tried to tackle this whole issue with platform responsibility
head on, and we appreciate the input we got from many. Last
Congress, we, as you heard, held hearings and legislated on
online platforms not fulfilling their Good Samaritan
obligations, especially when it comes to online human
trafficking.
Companies' use of algorithms and the impact such algorithms
have on influencing consumer behavior, we took a look at that.
Improving/expanding the reach of broadband services so rural
and urban consumers of all ages can benefit in a connected
world from the positive aspects of the internet. Explaining the
online advertising ecosystem, preservation and promotion across
border data flows, a topic we need to continue to work on.
Other related issues we face in the connected world, such as
cybersecurity, Internet of Things, artificial intelligence, to
name just a few.
We also invited the heads of the tech industry to come and
explain their practices right in this hearing room. Two of the
committee's highest-profile hearings in recent memory focused
squarely on platform responsibility. The CEO of Facebook, Mark
Zuckerberg, came and spent about 5\1/2\ hours right at that
table to answer some pretty tough questions on the Cambridge
Analytica debacle as well as provide the committee with more
insight into how Facebook collects consumer information and
what Facebook does with that information.
We also welcomed the CEO of Twitter, Jack Dorsey, to
provide the committee with more insight into how Twitter
operates, decisions Twitter makes on its platform, and how such
decisions impact consumers specifically, so voices don't feel
silenced.
I am pleased that Chairman Pallone brought in the CEO of
Reddit last year, and hope the trend will continue as we
understand this ever-evolving and critically important
ecosystem from those that sit on the top of it.
This hearing today helps with that, as this group of
experts shine a light on questionable practices I hope can
yield further fruitful results. Such efforts often lead to
swifter actions than any government action can get done.
Following our series of hearings, there is proof that some
companies are cleaning up their platforms, and we appreciate
the work you are doing. For example, following our hearing on
Cambridge Analytica, Facebook made significant changes to its
privacy policies and Facebook reformatted its privacy settings,
to make more accessible and user-friendly, ease the ability for
its users to delete and control their information, took down
malicious entities on its platform, and invested in programs to
preserve and promote legitimate local news operations.
And during that hearing, Representative McKinley actually
pushed Mr. Zuckerberg pretty hard on some specific ads he had
seen illegally selling opioids without prescriptions on
Facebook, and as a result, Facebook removed those ads. In fact,
we got a call, I think as Mr. Zuckerberg was headed to the
airport that afternoon, that those had already been taken down.
Also notable, through the Global Internet Forum to Counter
Terrorism, platforms such as Facebook, Twitter, and YouTube
have been working together to tackle terrorist content and,
importantly, disrupt violent extremists' ability to promote
themselves, share propaganda, and exploit digital platforms.
And we thank you for that work.
Now, this is not to suggest the online ecosystem is
perfect. It is far from it. Can these companies be doing more
to clean up their platforms? Of course, and I expect them to,
and I think you are all working on that.
So let me be very clear. This hearing should serve as an
important reminder to all online platforms that we are watching
them closely. We want to ensure we do not harm innovation, but,
as we have demonstrated in a bipartisan fashion in the past,
when we see issues or identify clear harms to consumers and we
do not see online entities taking appropriate action, we are
prepared to act.
So, Madam Chair, thanks for having this hearing. This is
tough stuff. I have a degree in journalism. I am a big advocate
of the First Amendment. And it can be messy business to, on the
one hand, call on them to take down things we don't like and
still stay on the right side of the First Amendment, because
vigorous speech, even when it is inaccurate, is still protected
under the First Amendment. And if you go too far, then we yell
at you for taking things down that we liked. And if you don't
take down things we don't like, then we yell at you for that.
So you are kind of in a bit of a box, and yet we know 230 is an
issue we need to revise and take a look at as well.
And then speaking of revise, I had to chuckle that we all
get the opportunity to revise and extend our remarks throughout
this process and clean up our bad grammar. So maybe some of
what we have is kind of fake reporting, but anyway, we will
leave that for another discussion on another day.
And, with that, I yield back, Madam Chair.
[The prepared statement of Mr. Walden follows:]
Prepared Statement of Hon. Greg Walden
Good morning, and welcome to our witnesses. I want to first
thank Chair Schakowsky for organizing today's incredibly
insightful hearing--which is focused on deception online.
For many years, the internet has been a force for good. It
provides consumers with unbelievable access to unlimited
information, goods and services, and people--no matter where
they are in the world.
But, as with anything, the internet presents bad actors and
those seeking to harm others ample opportunities to manipulate
users and take advantage of consumers, which often tend to be
some of our most vulnerable populations. Arguably, the digital
ecosystem is such that harmful acts are easily exacerbated and,
as we all know, false information or fake videos spread at
breakneck speeds. That is why when I was chairman of this
committee, we tackled platform responsibility head-on.
Last Congress, we held hearings and legislated on:
Online platforms not fulfilling their ``Good
Samaritan'' obligations, especially when it comes to online
human sex trafficking.
Companies' use of algorithms and the impact such
algorithms have on influencing consumer behavior;
Improving and expanding the reach of broadband
services so rural and urban, consumers of all ages, can benefit
in a connected world;
Explaining the online advertising ecosystem;
Preservation and promotion of cross-border data
flows; and
Other related issues we face in the connected
world such as cybersecurity, Internet of Things, artificial
intelligence, to name just a few.
We also invited the heads of tech industry to come explain
their practices in this hearing room. Two of the committee's
highest profile hearings in recent memory were focused squarely
on platform responsibility.
I brought in the CEO of Facebook, Mark Zuckerberg, to
answer tough questions on the Cambridge Analytica debacle, as
well as provide the committee with more insight into how
Facebook collects consumer information, and what Facebook does
with that information.
I also welcomed the CEO of Twitter, Jack Dorsey, to provide
the committee with more insight into how Twitter operates,
decisions Twitter makes on its platform, and how such decisions
impact consumers, specifically so voices don't feel silenced.
I am pleased that Chairman Pallone brought in the CEO of
Reddit last year and hope the trend will continue as we
understand this ever-evolving ecosystem from those that sit on
top of it. This hearing today helps with that as this group of
experts shine a light on questionable practices that I hope can
yield further fruitful results. Such efforts often lead to
swifter action than any government action can.
Following our series of hearings, there is proof that some
companies are cleaning up their platforms. For example,
following our hearing on the Cambridge Analytica scandal,
Facebook made significant changes to its privacy policies.
Facebook reformatted its privacy settings to make it more
accessible and user friendly; eased the ability for its users
to control and delete their information; took down malicious
entities on its platform; and, invested in programs to preserve
and promote legitimate local news operations. And during that
hearing Rep. McKinley pushed Mr. Zuckerberg on specific ads
he'd seen illegally selling opioids without prescription on
Facebook. As a result, Facebook removed the ads.
Also notable--through the Global Internet Forum to Counter
Terrorism--platforms such as Facebook, Twitter, and YouTube
have been working together to tackle terrorist content and,
importantly, disrupt violent extremists' ability to promote
themselves, share propaganda, and exploit digital platforms.
Now this is not to suggest the online ecosystem is
perfect--it is far from it. Can these companies be doing more
to clean up their platforms? Of course, they can, and I expect
them to.
So, let me be very clear: This hearing should serve as an
important reminder to all online platforms that we are watching
them closely. We want to ensure we do not harm innovation, but
as we have demonstrated in a bipartisan fashion in the past,
when we see issues or identify clear harms to consumers and we
do not see online entities taking appropriate action, we are
prepared to act.
Thank you. I yield back.
Ms. Schakowsky. The gentleman yields back.
And the Chair would like to remind Members that, pursuant
to committee rules, all Members' opening statements shall be
made part of the record.
I would now like to introduce our witnesses for today's
hearing.
Ms. Monika Bickert, vice president of Global Policy
Management at Facebook. I want to acknowledge and thank you,
Ms. Bickert. I know that you are not feeling well today and may
want to abbreviate some of your testimony, but we thank you
very much for coming anyway.
I want to introduce Dr. Joan Donovan, research director of
the Technology and Social Change Project at the Shorenstein
Center on Media, Politics and Public Policy at Harvard Kennedy
School.
Mr. Justin Hurwitz, assistant professor of law and director
of NU Governance and Technology Center at the University of
Nebraska College of Law, and director of law and economics
programs at the International Center for Law and Economics.
And finally, Dr. Tristan Harris, who is executive director
for the Center for Humane Technology.
We want to thank our witnesses for joining us today. We
look forward to your testimony.
At this time, the Chair will recognize each witness for 5
minutes to provide their opening statement. Before we begin, I
would just like to explain the lighting system for those who
may not know it. In front of you are a series of lights. The
lights will initially be green at the start of your opening
statement. The light will turn to yellow when you have 1 minute
remaining, and if you could please begin to wrap up your
testimony at that point, and then the light will turn red when
your time has expired.
So, Ms. Bickert, you are recognized for 5 minutes.
STATEMENTS OF MONIKA BICKERT VICE PRESIDENT OF GLOBAL POLICY
MANAGEMENT, FACEBOOK; JOAN DONOVAN, Ph.D., DIRECTOR, TECHNOLOGY
AND SOCIAL CHANGE PROJECT, SHORENSTEIN CENTER ON MEDIA,
POLITICS AND PUBLIC POLICY, HARVARD KENNEDY SCHOOL; JUSTIN
(GUS) HURWITZ, DIRECTOR OF LAW AND ECONOMICS PROGRAMS,
INTERNATIONAL CENTER FOR LAW AND ECONOMICS; AND TRISTAN HARRIS,
PRESIDENT AND COFOUNDER, CENTER FOR HUMANE TECHNOLOGY
STATEMENT OF MONIKA BICKERT
Ms. Bickert. Thank you, Chairwoman Schakowsky, Ranking
Member McMorris Rodgers, and other distinguished members of the
subcommittee. Thank you for the opportunity to appear before
you today.
My name is Monika Bickert. I am the vice president for
Global Policy Management at Facebook, and I am responsible for
our content policies. As the chairwoman pointed out, I am a
little under the weather today so, with apologies, I am going
to keep my remarks short, but will rely on the written
testimony I have submitted.
We know that we have an important role to play at Facebook
in addressing manipulation and deception on our platform. And
we have many aspects to our approach, including our community
standards, which specify what we will remove from the site, and
our relationship with third-party fact checkers, through which
fact-checking organizations can rate content as false. We put a
label over that content saying that this is false information,
and we reduce its distribution.
Under the community standards, there are some types of
misinformation that we remove, such as attempts to suppress the
vote or to interfere with the Census. And we announced
yesterday a new prong in our policy where we will also remove
videos that are edited or synthesized, using artificial
intelligence, or deep learning techniques, in ways that are not
apparent to the average person that would mislead the average
person to believe that the subject of the video said something
that he or she did not, in fact, say.
To be clear, manipulated media that doesn't fall under this
new policy definition is still subject to our other policies
and our third-party fact checking. That means that deepfakes
are still an emerging technology. One area where internet
experts have seen them is in nudity and pornography. All of
that violates our policies against nudity and pornography, and
we would remove it. Manipulated videos are also eligible to be
fact-checked by these third-party fact-checking organizations
that we work with to label and reduce the distribution of
misinformation.
We are always improving our policies and our enforcement,
and we will continue to do the engagement we have done outside
the company with academics and experts to understand the new
ways that these technologies are emerging and affecting our
community. We would also welcome the opportunity to collaborate
with other industry partners and interested stakeholders,
including academics, civil society, and lawmakers, to help
develop a consistent industry approach to these issues. Our
hope is that by working together with all of these
stakeholders, we can make faster progress in ways that benefit
all of society.
Thank you, and I look forward to your questions.
[The prepared statement of Ms. Bickert follows:]
[GRAPHICS NOT AVAILABLE IN TIFF FORMAT]
Ms. Schakowsky. Thank you.
And now, Dr. Donovan, you are recognized for 5 minutes.
STATEMENT OF JOAN DONOVAN, Ph.D.
Dr. Donovan. Thank you, Chairwoman Schakowsky, Ranking
Member McMorris Rodgers, Chairman Pallone, and Ranking Member
Walden, for having me today. It is truly an honor to be
invited.
I lead a team at Harvard Kennedy's Shorenstein Center that
researches online manipulation and deception, and I have been a
researcher of the internet for the last decade. So I know quite
a bit about changes in policies as well as the development of
platforms themselves and what they were intended to do.
One of the things that I want to discuss today is online
fraud, which is a great deal more widespread than many
understand. Beyond malware, spam, and phishing attacks, beyond
credit card scams and product knock-offs, there is a growing
threat from new forms of identity fraud enabled by
technological design. Platform companies are unable to manage
this alone, and Americans need governance. Deception is now a
multimillion-dollar industry.
My research team tracks dangerous individuals and groups
who use social media to pose as political campaigns, social
movements, news organizations, charities, brands and even
average people. This emerging economy of misinformation is a
threat to national security. Silicon Valley corporations are
largely profiting from it, while key political and social
institutions are struggling to win back the public's trust.
Platforms have done more than just given users a voice
online. They have effectively given them the equivalent of
their own broadcast station, emboldening the most malicious
among us. To wreak havoc with a media manipulation campaign,
all one bad actor needs is motivation. Money also helps. But
that is enough to create chaos and divert significant resources
from civil society, politicians, newsrooms, healthcare
providers, and even law enforcement, who are tasked with
repairing the damage. We currently do not know the true cost of
misinformation.
Individuals and groups can quickly weaponize social media,
causing others financial and physical injury. For example,
fraudsters using President Trump's image, name, logo and voice
have siphoned millions from his supporters by claiming to be
part of his reelection coalition. In an election year,
disinformation and donation scams should be of concern to
everyone. Along with my coresearchers Brian Friedberg and
Brandi Collins-Dexter, I have studied malicious groups,
particularly white supremacists and foreign actors, who have
used social media to inflame racial divisions. Even as these
imposters are quickly identified by the communities they
target, it takes time for platforms to remove inciting content.
A single manipulation campaign can create an incredible strain
on breaking news cycles, effectively turning many journalists
into unpaid content moderators and drawing law enforcement
towards false leads.
Today, I argue that online communication technologies need
regulatory guardrails to prevent them from being used for
manipulative purposes. And in my written testimony, I have
provided a longer list of ways that you could think about
technology differently.
But right now, I would like to call attention to
deceptively edited audio and video to drive clicks, likes, and
shares. This is the AI technology commonly referred to as
deepfakes. And what I would also like to point out, with my
coresearcher Britt Paris, that we have argued that cheapfakes
are a wider threat. Like the doctored video of Speaker Pelosi,
last week's decontextualized video of Joe Biden seemingly
endorsing a white supremacist talking point poses another
substantial challenge. Because the Biden video was clipped from
nonaugmented footage, platforms refused to take down this
cheapfake. Millions have now seen it.
Platforms, like radio towers, provide amplification power
and, as such, they have a public-interest obligation. And I
point out here that platforms are highly centralized mechanisms
of distribution, while the internet is not. So I am not trying
to conflate platforms with the internet, but this is why we
place the burden of moderation on platforms and not with ISPs.
The world online is the real world, and this crisis of
counterfeits threatens to disrupt the way Americans live our
lives. Right now, malicious actors jeopardize how we make
informed decisions about who to vote for and what causes we
support, while platform companies have designed systems that
facilitate this manipulation.
We must expand the public understanding of technology by
guarding consumer rights against technological abuse, including
a cross-sector effort to curb the distribution of harmful and
malicious content. As Danah Boyd and I have written, platform
companies must address the power of amplification and
distribution separately from content, so that media
distribution is transparent and accountable. I urge Congress to
do the same. Platforms and politics and regulation and
technology must work in tandem, or else the future is forgery.
Thank you.
[The prepared statement of Dr. Donovan follows:]
[GRAPHICS NOT AVAILABLE IN TIFF FORMAT]
Ms. Schakowsky. Thank you.
And now, Mr. Hurwitz, you are recognized for 5 minutes.
STATEMENT OF JUSTIN (GUS) HURWITZ
Mr. Hurwitz. Thank you, Ms. Chairwoman, along with members
of the committee, for the opportunity to speak to you today. I
would also be remiss if I did not thank my colleague Kristian
Stout and research assistant Justin McCully for help in
drafting my written testimony.
I am a law professor, so I apologize. I will turn to
discussing the short law review article I have written for you
as my testimony and assigned to you to read in a moment. Before
I turn to that, I want to make a couple of book
recommendations. If you really want to understand what is at
stake with dark patterns, you should start by reading Brett
Frischmann and Evan Selinger's recent book, ``Re-Engineering
Humanity.'' In my spare time, I am a door-to-door book
salesman. I have a copy here. Their book discusses how modern
technology, data analytics, combined with highly programmable
environments, are creating a world in which people are, to use
their term, programmable. This book will scare you.
After you read that book, you should then read Cliff Kuang
and Robert Fabricant's recent book, ``User Friendly.'' This was
just published in November. It discusses the importance and
difficulty of designing technologies that seamlessly operate in
line with user expectations as user-friendly technologies. This
book will help you understand the incredible power of user-
friendly design and fill you with hope for what design makes
possible, along with appreciation for how difficult it is to do
design well. Together, these books will show you both sides of
the coin.
Dark patterns are something that this committee absolutely
should be concerned about, but this committee should also
approach the topic with great caution. Design is powerful, but
it is incredibly difficult to do well. Efforts to regulate bad
uses of design could easily harm efforts to do and use design
for good.
How is that for having a professor testify? I have already
assigned two books and a law review article of my own for you
to read. I will do what I can to summarize some of the key
ideas from that article in the next 3 minutes or so.
Dark pattern is an ominous term. It is itself a dark
pattern. It is a term for a simple concept. People behave in
predictable ways. These behavioral patterns can be used to
program us in certain ways, and the concern is that sometimes
we can be programmed to act against our own self-interest.
So I have some examples. If we can look at the first
example, this is something from the internet.
[Slide shown, included in Mr. Hurwitz's prepared statement
below.]
You look at this for a moment. Who here feels manipulated
by this image? It is OK to say yes. I do. The designer of this
image is using his knowledge of how people read text in an
image to make it feel like the image is controlling us, making
us control how our eyes are following it and predicting where
we are going to go next. Weird stuff.
Let's look at another example. Again, you can definitely
tell from the internet.
[Slide shown, included in Mr. Hurwitz's prepared statement
below.]
Again, who feels like this image is manipulative? The
previous image was harmless, but this one hints at the darker
power of dark patterns. Most of you probably missed the typos
in the first line and then the second line until the text
points them out to you. What if this had been a contract and
this trick was used to insert a material term or distract you
from a material term in the contract that you were agreeing to?
This has now gone from weird stuff to scary stuff.
On the other hand, these same tricks can be used for good.
In this same example, what if this trick were used to highlight
an easily missed but important concern for consumers to pay
attention to? This could be beneficial to consumers.
Design is not mere aesthetics. All design influences how
designs are made. It is not possible to regulate bad design
without also affecting good design.
So how much of a problem are dark patterns? Recent research
shows that websites absolutely are using them, sometimes
subtly, sometimes overtly, to influence users. And other
research shows us that these tactics can be effective, leading
consumers to do things that they otherwise wouldn't do. We have
already heard some examples of these, so I won't repeat what
has already been discussed. Rather, I would like to leave you
with a few ideas about what, if anything, we should do about
them.
First, dark patterns are used both online and offline.
Stores use their floor plans to influence what people buy.
Advertisers make consumers feel a sense of need and urgency for
products. Try canceling a subscription service or returning a
product. You will likely be routed through a maddening maze of
consumer service representatives. If these patterns are a
problem online, they are a problem offline, too. We shouldn't
focus on one to the exclusion of the other.
Second, while these tricks are annoying, it is unclear how
much they actually harm consumers or how much benefit they may
confer. Studies of mandatory disclosure laws, for instance,
find that they have limited effectiveness. On the other hand,
these tricks can also be used to benefit consumers. We should
be cautious with regulations that may fail to stop bad conduct
while reducing the benefits of good conduct.
Third, most of the worst examples of dark patterns very
likely fall within the FTC's authority to regulate deceptive
acts or practices. Before the legislature takes any action to
address these concerns, the FTC should attempt to use its
existing authority to address them. It is already having
hearings on these issues. If this proves ineffective, the FTC
should report to you, to Congress, on these practices.
Fourth, industry has been responsive to these issues and,
to some extent, has been self-regulating. Web browsers and
operating systems have made many bad design practices harder to
use. Design professionals scorn dark patterns practices.
Industry standardization and best practices and self-
regulations should be encouraged.
Fifth, regulators should----
Ms. Schakowsky. Wrap it up.
Mr. Hurwitz. Yes. Last and building on all of the above,
this is an area well-suited to cooperation between industry and
regulators. Efforts at self-regulation should be encouraged and
rewarded. Perhaps even more important, given the complexity of
these systems, industry should be at the front line of
combating them. Industry has greater design expertise and
ability to experiment than regulators, but there is an
important role for regulation to step in where industry fails
to police itself.
In a true professor--thank you. I look forward to
discussion.
[The statement of Mr. Hurwitz follows:]
[GRAPHICS NOT AVAILABLE IN TIFF FORMAT]
Ms. Schakowsky. So, Mr. Harris, you are recognized now for
5 minutes.
STATEMENT OF TRISTAN HARRIS
Mr. Harris. Thank you, Chairwoman Schakowsky and members. I
really appreciate you inviting me here.
I am going to go off script. I come here because I am
incredibly concerned. I actually have a lifelong experience
with deception and how technology influences people's minds. I
was a magician as a kid, so I have started off by seeing the
world this way. And then I studied at a lab called the Stanford
Persuasive Technology Lab, actually with the founders of
Instagram. And so I know the culture of the people who build
these products and the way that it is designed intentionally
for mass deception.
I think there is--the thing I most want to respond to here
is we often frame these issues as we have got a few bad apples.
We have got these bad deepfakes, we have got to get them off
the platform. We have got this bad content. We have got these
bad bots. What I want to argue is this is actually--and we have
got these dark patterns.
What I want to argue is we have dark infrastructure. This
is now the infrastructure by which 2.7 billion people, bigger
than the size of Christianity, make sense of the world. It is
the information environment. And if someone went along, private
companies, and built nuclear power plants all across the United
States, and they started melting down and they said, ``Well, it
is your responsibility to have HazMat suits and, you know, have
a radiation kit,'' that is essentially what we are experiencing
now. The responsibility is being put on consumers when, in
fact, if it is the infrastructure, it should be put on the
people building that infrastructure.
There are specifically two areas of harm I want to focus
on, even though when this becomes the infrastructure it
controls all of our lives. So we wake up with these devices. We
check our phones 150 times a day. It is the infrastructure for
going to bed. Children spend as much time on these devices as
they do at the hours at school. So no matter what you are
putting in people's brains, kids' brains at school, you have
got all the hours they spend, you know, on their phones.
And let's take the kids' issue. So as infrastructure, the
business model of this infrastructure is not aligned with the
fabric of society. How much have you paid for your Facebook
account recently, or your YouTube account? Zero. How are they
worth more than a trillion dollars in market value? They
monetize our attention. The way they get that attention is by
influencing you and using the dark patterns or tricks to do it.
So the way they do it with children is they say, ``How many
likes or followers do you have?'' So they basically get
children addicted to getting attention from other people. They
use filters, likes, et cetera, beautification filters that
enhance your self-image. And after two decades in decline, the
mental health of teen girls, high-depressive symptoms--there is
an image here that they will be able to show--went up 170
percent after the year 2010, with the rise of Instagram, et
cetera. OK. These are your children. These are your
constituents. This is a real issue. It is because we are
hacking the self-image of children.
On the information ecology front, the business model, think
of it like we are drinking from the Flint water supply of
information. The business model is polarization, because the
whole point is I have to figure out and calculate whatever
keeps your attention, which means affirmation, not information,
by default. It polarizes us by default.
There is a recent Upturn study that it actually costs more
money to advertise across the aisle than it does to advertise
to people with your own same beliefs. In other words,
polarization has a home field advantage in terms of the
business model. The natural function of these platforms is to
reward conspiracy theories, outrage, what we call the race to
the bottom of the brainstem. It is the reason why all of you at
home have crazier and crazier constituents who believe crazier
and crazier things, and you have to respond to them. I know you
don't like that.
Russia is manipulating our veterans by--we have totally
open borders. While we have been protecting our physical
borders, we left the digital border wide open. Imagine a
nuclear plant and you said we are not going to actually protect
the nuclear plants from Russian cyber attacks. Well, this is
sort of like Facebook building the information infrastructure
and not protecting it from any bad actors until that pressure
is there.
And this is leading to a kind of information trust
meltdown, because no one even has to use deepfakes for
essentially people to say, ``Well, that must be a faked video,
right?'' So we are actually at the last turning point, kind of
an event horizon, where we either protect the foundations of
our information and trust environment or we let it go away.
And, you know, we say we care about kids' education, but we
allow, you know, technology companies to basically tell them
that the world revolves around likes, clicks, and shares. We
say we want to, you know, come together, but we allow
technology to profit by dividing us into echo chambers. We say
America should lead on the global stage against China with its
strong economy, but we allow technology companies to degrade
our productivity and mental health, while jeopardizing the
development of our future workforce, which is our children.
And so, while I am finishing up here, I just want to say
that, instead of trying to design some new Federal agency, some
master agency, when technology has basically taken all the laws
of the physical world--taken all the infrastructure of the
physical world and virtualized it into a virtual world with no
laws--what happens when you have no laws for an entire
virtualized infrastructure? You can't just bring some new
agency around and regulate all of the virtual world.
Why don't we take the existing infrastructure, existing
agencies who already have purview--Department of Education,
Health and Human Services, Natural Institutes of Health--and
have a digital update that expands their jurisdiction to just
ask, well, how do we protect the tech platforms in the same
areas of jurisdiction?
I know I am out of time, so thank you very much.
[The statement of Mr. Harris follows:]
[GRAPHICS NOT AVAILABLE IN TIFF FORMAT]
Ms. Schakowsky. Thank you.
So now we have concluded our witnesses' opening statements.
At this time, we will move to Member questions. Each Member
will have 5 minutes to ask a question of our witnesses. I will
begin by recognizing myself for 5 minutes.
So, as chair of the subcommittee, over and over again I am
confronted with new evidence that Big Tech has failed in
regulating itself. When we had Mark Zuckerberg here, I kind of
did a review of all the apologies that we have had from him
over the years, and I am concerned that Facebook's latest
effort to address misinformation on the platforms leaves a lot
out.
I want to begin with some questions of you, Ms. Bickert. So
the deepfakes policy only covers video, as I understand it,
that has been manipulated using artificial intelligence, or
deep learning. Is that correct?
Ms. Bickert. Thank you, Chairwoman Schakowsky. The policy
that we announced yesterday is confined to the definition that
we set forth about artificial intelligence being used in a
video to make it appear that somebody is saying something----
Ms. Schakowsky. I only have 5 minutes. So the video, for
example, of Speaker Pelosi was edited to make her look like she
was drunk, wouldn't have been taken down under the new policy.
Is that right, yes or no?
Ms. Bickert. It would not fall under that policy, but it
would still be subject to our other policies that address
misinformation.
Ms. Schakowsky. And, as I read the deepfakes policy, it
only covers video where a person is made to appear like they
said words that they didn't actually say, but it doesn't cover
videos where just the image is altered. Is that true?
Ms. Bickert. Chairwoman Schakowsky, that is correct about
that policy. We do have a broader approach to misinformation
that would put a label--we would actually obscure the image and
put a screen over it that says ``false information,'' and
directs people to information from fact checkers.
Ms. Schakowsky. So, Ms. Bickert, I really don't understand
why Facebook should treat fake audio differently from fake
images. Both can be highly misleading and result in significant
harm to individuals and undermine democratic institutions.
Dr. Donovan, in your testimony, you noted that, quote,
``cheapfakes,'' unquote, are more prevalent than deepfakes. Do
you see any reason to treat deepfakes and cheapfakes
differently?
Dr. Donovan. One of the things----
Ms. Schakowsky. Microphone.
Dr. Donovan. Of course, as if I am not loud enough.
One of the things that cheapfakes leverage is what is sort
of great about social media, is that it makes things clippier,
or smaller. And so I understand the need for separate policies,
but also the cheapfakes issue has not been enforced. Speaking
more broadly about social media platforms in general, there is
completely uneven enforcement.
So you can still find that piece of misinformation within
the wrong context in multiple places. And so the policy on
deepfakes is both narrow--and I understand why--but also, one
thing that we should understand is presently there is no
consistent detection mechanism for even finding deepfakes at
this point. And so I would be interested to know more about how
they are going to seek out, either on upload, not just
Facebook----
Ms. Schakowsky. I am going to have to cut you off at this
point, because I do want to ask Mr. Harris.
Given the prevalence of deceptive content online, are
platforms doing enough to stop the dissemination of
misinformation, and what can government do to prevent such
manipulation of consumers? Should government be seeking to
clarify the principle that if it is illegal offline then it is
illegal online?
Mr. Harris. Yes. A good example of that--so first is no,
the platforms are not doing enough, and it is because their
entire business model is misaligned with solving the problem.
And I don't vilify the people because of that. It is just their
business model is against the issue.
We used to have Saturday morning cartoons. We protected
children from certain kinds of advertising, time/place/manner
restrictions. When YouTube gobbles up that part of the
attention economy, we lose all those protections. So why not
bring back the protections of Saturday morning? We used to have
fair-price/equal-price election ads on TV, the same price for
each politician to reach someone. When Facebook gobbles up
election advertising, we just removed all of those same
protections.
So we are basically moving from a lawful society to an
unlawful virtual internet society, and that is what we have to
change.
Ms. Schakowsky. Thank you. I yield back.
And now the Chair recognizes Mrs. Rodgers, our subcommittee
ranking member, for 5 minutes.
Mrs. Rodgers. Thank you, Madam Chair.
I referenced how misinformation is not a new problem, but
certainly with the speed of information, how it can travel in
the online world, its harm is increasing. That said, I have
long believed that the way to address information is more
transparency, more sources, more speech, not less. This is
important, not just in an election cycle, but also in
discussions around public health issues, natural disasters, or
any number of significant events. I am worried about this
renewed trend, where some want the government to set the
parameters and potentially limit speech and expression.
Ms. Bickert, how does free speech and expression factor
into Facebook's content decisions, and can you please explain
your use of third-party fact checkers?
Ms. Bickert. Thank you. We are very much a platform for
free expression. It is one of the reasons that we work with
third-party fact-checking organizations, because what we do if
they have ranked something false is, we share more information
on the service. So we put a label over it, this is false
information, but then we show people here is what fact checkers
are saying about this story.
We work with more than 50 organizations worldwide, and
those organizations are chosen after meeting high standards for
fact checking.
Mrs. Rodgers. Thank you. As a followup, with the total
volume of traffic you have, clearly human eyes alone can't keep
up. So artificial intelligence and machine learning have a
significant role to identify not only deepfakes but also other
content that violates your terms of service. Would you just
explain a little bit more to us how you use AI and the
potential to use AI to fight fire with fire?
Ms. Bickert. Absolutely. We do use a combination of
technology, and people to identify potential information to
send to fact checkers. We also use people and technology to try
to assess whether or not something has been manipulated, media.
That would be covered by the policy we released yesterday.
So, with the fact-checking program, we use technology to
look for things like--let's say somebody has shared an image or
a news story and people are--friends are commenting on that,
saying, ``Don't you know this is a hoax?'' or ``This isn't
true.'' That is the sort of thing our technology can spot and
send that content over to fact-checkers.
But it is not just technology. We also have ways for people
to flag if they are seeing something that they believe to be
false. That can send content over to fact checkers. And then
the fact checkers can also proactively choose to rate something
that they are seeing on Facebook.
Mrs. Rodgers. Thank you.
Professor Hurwitz, can you briefly describe how user
interfaces can be designed to shape consumer choice and how
such designs may benefit or harm consumers?
Mr. Hurwitz. They can be used--they can be modified,
created, structured in any number of ways. We have heard
examples: font size, text placement, the course of interaction
with a website, or even just a phone menu system. These can be
used to guide users into making uninformed decisions, or to
highlight information that users should be paying attention to.
This broadly falls into the category of nudges and behavioral
psychology. That is an intensely researched area. It can be
used in many ways.
Mrs. Rodgers. You highlighted some of that in your
testimony. Would you explain how the FTC can use its existing
Section 5 authority to address most of the concerns raised by
dark pattern practices?
Mr. Hurwitz. Yes, very briefly. I could lecture for a
semester on this, not to say that I have.
The FTC has a broad history, long history of regulating
unfair and deceptive practices and advertising practices. Its
deception authority--false statements, statements that are
material to a consumer, making a decision that is harmful to
the consumer. They can use adjudication. They can enact rules
in order to take action against platforms or any entity, online
or offline, that deceives consumers.
Mrs. Rodgers. Do you think that they are doing enough?
Mr. Hurwitz. I would love to see the FTC do more in this
area, especially when it comes to rulemaking and in-court
enforcement actions, because the boundaries of their authority
are unknown, uncertain, untested. This is an area where
bringing suits, bringing litigation, that tells us what the
agency is capable of, which this body needs to know before it
tries to craft more legislation or give more authority to an
entity. If we already have an agency that has power, let's see
what it is capable of.
Mrs. Rodgers. Right. OK. Thank you, everyone. I appreciate
you all being here. Very important subject, and I appreciate
the Chair for hosting, or having this hearing today.
Ms. Schakowsky. I thank the ranking member, who yields
back. And now I recognize the chair of the full committee, Mr.
Pallone, for 5 minutes.
Mr. Pallone. Thank you, Madam Chair.
I have got a lot to ask here, so I am going to ask you for
your responses to be brief, if possible. But, in your various
testimonies, you all talked about a variety of technologies and
techniques that are being used to deceive and manipulate
consumers.
We have heard about user interfaces designed to persuade
and sometimes trick people into making certain choices,
deepfakes and cheapfakes, that show fictional scenarios that
look real, and algorithms designed to keep people's eyes locked
on their screens. And we know these things are happening. But
what is less clear is how and the extent to which these
techniques are being used commercially and on commercial
platforms.
So first let me ask Dr. Donovan: As a researcher who
focuses on the use of these techniques, do you have sufficient
access to commercial platform data to have a comprehensive
understanding of how disinformation and fraud is conducted and
by whom?
Dr. Donovan. The brief answer is no, and that is because we
don't have access to the data as it is. There are all these
limits on the ways in which you can acquire data through the
interface.
And then the other problem is that there was a very good-
faith effort between Facebook and scholars to try to get a
bunch of data related to the 2016 election. That fell apart,
but a lot of people put an incredible amount of time, money,
and energy into that effort, and it failed around the issues
related to privacy and differential privacy.
What I would love to see also happen is, Twitter has
started to give data related to deletions and account
takedowns. We need a record of that so that, when we do audit
these platforms for either financial or social harms, that the
deletions are also included and marked. Because, even if you
can act like a data scavenger and go back and get data, when
things are deleted, sometimes they are just gone for good, and
those pieces of information are often the most crucial.
Mr. Pallone. Thank you.
Mr. Harris, should the government be collecting more
information about such practices in order to determine how best
to protect Americans?
Mr. Harris. Yes. Here is an example: So, unlike other
addictive industries, for example--addiction is part of the
deception that is going on here--the tobacco industry doesn't
know which users are addicted to smoking, the alcohol industry
doesn't know exactly who is addicted to alcohol. But, unlike
that, each tech company does know exactly how many people are
checking more than, you know, 100 times a day between certain
ages. They know who is using it late at night.
And you can imagine using existing agencies--say,
Department of Health and Human Services--to be able to audit
Facebook on a quarterly basis and say, ``Hey, tell us how many
users are addicted between these ages, and then what are you
doing next quarter to make adjustments to reduce that number?''
And every day they are the ones issuing the questions, and the
responsibility and the resources have to be deployed by the
actor that has the most of them, which in this case would be
Facebook. And there is a quarterly loop between each agency
asking questions like that, forcing accountability with the
companies for the areas of their existing jurisdiction.
So I am just trying to figure out is that a way that we can
scale this to meet the scope of the problem. You realize this
is happening to 2.7 billion people.
Mr. Pallone. Thank you. This week, Facebook released a new
policy on how it will handle deepfakes. So, Ms. Bickert, under
your policy deepfakes are--and I am paraphrasing--videos
manipulated through artificial intelligence that are intended
to mislead and are not parody or satire. Did I get that right?
Ms. Bickert. Yes, that is right.
Mr. Pallone. OK. Now, I understand that Twitter and YouTube
either do not have or use the same definition for deepfakes,
and that is indicative of a lack of consistent treatment of
problematic content across the major platforms. Banned hate
speech or abusive behavior on one site is permitted on another.
There seems to be very little consistency across the
marketplace, which leaves consumers at a loss.
So let me go to Dr. Donovan again. Is there a way to
develop a common set of standards for these problematic
practices so that consumers are not facing different policies
on different websites?
Dr. Donovan. I think it is possible to create a set of
policies, but you have to look at the features that are
consistent across these platforms. If they do, for instance,
use attention to a specific post in their algorithms to boost
popularity, then we need a regulation around that, especially
because bots or unmanned accounts, for lack of a better term,
are often used to accelerate content and to move content across
platforms.
These are things that are usually purchased off-platform,
and they are considered a dark market product, but you can
purchase attention to an issue. And so, as a result, there has
to be something more broad that goes across platforms, but also
looks at the features and then also tries to regulate some of
these markets that are not built into the platform themselves.
Mr. Pallone. All right. Thank you.
Thank you, Madam Chair.
Ms. Schakowsky. Thank you.
Mr. Bucshon, you are recognized for 5 minutes.
Mr. Bucshon. Thank you, Madam Chairwoman. I am sorry, I
have two of these hearings going on at the same time, so I am
back and forth.
I appreciate the hearing and the opportunity to discuss the
spread of misinformation on the internet, but I want to stress
that I am concerned over the efforts to make tech companies the
adjudicators of ``truth,'' in quotation marks.
In a country founded on free speech, we should not be
allowing private corporations, in my view, or, for that matter,
the government to determine what qualifies as, again in
quotation marks, the ``truth,'' potentially censoring a voice
because that voice disagrees with a mainstream opinion. That
said, I totally understand the difficulty and the challenges
that we all face together concerning this issue, and how we
are, together, trying to work to address it.
Ms. Bickert, can you provide some more information on how
Facebook might or will determine if a video misleads? What
factors might you consider?
Ms. Bickert. Thank you. Just to be clear, there are two
ways that we might be looking at that issue. One is with regard
to the deepfakes policy that we released yesterday. And we will
be looking to see, specifically, were we seeing artificial
intelligence and deep learning? Was that part of the technology
that led to change or fabricate a video in a way that really
wouldn't be evident to the average person? And that will be a
fundamental part of determining whether there is misleading.
Separately----
Mr. Bucshon. Can I ask a question? Who is the average--
sorry, I will wait until you quit coughing so you can hear me.
Ms. Bickert. I am sorry.
Mr. Bucshon. The question then--I mean, I am playing
devil's advocate here--who is the average person?
Ms. Bickert. Congressman, these are exactly the questions
that we have been discussing with more than 50 experts as we
have tried to write this policy and get it in the right place.
Mr. Bucshon. And I appreciate what you are doing. I am not
trying to be difficult here.
Ms. Bickert. No, these are real challenging issues. It is
one of the reasons that we think, generally, the approach to
misinformation of getting more information out there from
accurate sources is effective.
Mr. Bucshon. And you stated in your testimony that, once a
fact checker rates a photo or video as false, or partly false,
Facebook reduces the distribution. Is there a way for an
individual who may have posted these things to protest the
decision?
Ms. Bickert. Yes, Congressman. They can go directly to the
fact checker. We make sure there is a mechanism for that. And
they can do that either if they dispute it or if they have
amended whatever it was in their article that was the problem.
Mr. Bucshon. Right. Because I would say--I mean, people
with good lawyers can dispute a lot of things, but the average
citizen in southwest Indiana who posts something online, there
needs to be, in my view, a fairly straightforward process that
the average person, whoever that might be, can understand to
protest or dispute the fact that their distribution has been
reduced. Thank you.
Mr. Hurwitz, you have discussed that the FTC has current
authority to address dark pattern. However, I would be
interested to know your thoughts on how consumers can protect
themselves from these patterns and advertisements. Is the only
solution through government action, or can consumer education
help highlight these advertisement practices?
Mr. Hurwitz. The most important thing for any company,
especially in the online context, is trust, the trust of the
consumers. Consumer education, user education, is important,
but I think that it is fair to say, with condolences perhaps to
Ms. Bickert, Facebook has a trust problem. If consumers--if
users stop trusting these platforms, if hearings such as this
shine a light on bad practices, then they are going to have a
hard time retaining users and consumers. That puts a great deal
of pressure.
In addition, stability of practices. One dark pattern is to
constantly change the user interface, so users don't know how
it operates. If we have stability, if we have platforms that
operate in consistent, predictable ways, that helps users
become educated, helps users understand what the practices are,
and learn how to operate in this new environment. Trust on the
internet is different. We are still learning what it means.
Mr. Bucshon. And I know you went over this, but can you
talk again about how these dark pattern practices took place
before the internet and are currently happening in brick-and-
mortar stores and other areas, mail pieces that politicians
send out.
I mean, I just want to reiterate again: This is a broader
problem than just the internet, this is something that has been
around for a while.
Mr. Hurwitz. Yes. Dark patterns, these practices, they go
back to the beginning of time. Fundamentally, they are
persuasion. If I want to convince you of my world view, if I
want to convince you to be my customer, if I want to convince
you to be my friend, I am going to do things that influence
you. I am going to present myself to you in ways that are going
to try and get you to like me or my product.
If you come into my store and ask for a recommendation--
``What size tire do I need for my car?''--my sales
representative is going to give you information. The store is
going to be structured--these have been used consistently
throughout----
Mr. Bucshon. My time is expired. My point was is that, when
we look at this problem, we need to, in my view, take a
holistic approach about what has happened in the past and, with
emerging technology, how we address that consistently and not
just target specific industries.
Thank you. I yield back.
Ms. Schakowsky. The gentleman yields back.
I now recognize Congresswoman Castor for 5 minutes.
Ms. Castor. Well, thank you, Chairwoman Schakowsky, for
calling this hearing.
You know, the internet and online platforms have developed
over time without a lot of safeguards for the public. And
government here, we exercise our responsibility to keep the
public safe, whether it is the cars we drive, or the water we
drink, airplanes, drugs that are for sale. And really, the same
should apply to the internet and online platforms.
You know, there is a lot of illegal activity being promoted
online, where the First Amendment just does not come into play.
And I hope we don't go down that rabbit hole, because we are
talking about human trafficking, terrorist plots, illicit sales
of firearms, child exploitation.
And now, what we have swamping these online platforms that
control the algorithms that manipulate the public are the
deepfakes, these dark patterns, artificial intelligence,
identity theft. But these online platforms, remember, they
control these algorithms that steer children and adults,
everyone in certain directions, and we have got to get a handle
on that.
For example, Mr. Harris, one manipulative design technique
is the autoplay feature. It is now ubiquitous across video
streaming platforms, particularly billions of people that go
onto YouTube or Facebook. This feature automatically begins
playing a new video after the current video ends. The next
video is determined using an algorithm. It is designed to keep
the viewer's attention.
This platform-driven algorithm often drives the
proliferation of illegal activities and dangerous ideologies
and conspiracy theories. It makes it much more difficult for
the average person to try to get truth-based content.
I am particularly concerned about the impact on kids, and
you have raised that and I appreciate that. You discuss how the
mental health of kids today really is at risk. Can you talk
more about the context in which children may be particularly
harmed by these addiction-maximizing algorithms and what
parents can do to protect kids from becoming trapped in a
YouTube vortex, and what you believe our responsibility is as
policymakers?
Mr. Harris. Thank you so much for your question. Yes, this
is very deeply concerning to me.
So laying it out, with more than 2 billion users, think of
these on YouTube as 2 billion ``Truman Shows.'' Each of you get
a channel, and a super computer is just trying to calculate the
perfect thing to confirm your view of reality. This, by
definition, fractures reality into 2 billion different
polarizing channels, each of which is tuned to bring you to a
more extreme view.
The quick example is, imagine a spectrum of all the videos
on YouTube laid out in one line, and on my left side over here,
you have the calm Walter Cronkite, rational science side of
YouTube, and the other side you have Crazy Town. You have UFOs,
conspiracy theories, Alex Jones, crazy stuff.
No matter where you start on YouTube, you could start in
the calm section or you could start in crazy. If I want you to
watch more, am I going to steer you that way or that way? I am
always going to steer you towards Crazy Town. So imagine taking
the ant colony of 2.1 billion humans and then just tilting it
like that.
Three examples of that per your kids example: 2 years ago
on YouTube, if a teen girl watched a dieting video, it would
autoplay anorexia videos, because those were more extreme. If
you watched a 9/11 news video, it would recommend 9/11
conspiracy theories. If you watched videos about the moon
landing, it would recommend flat Earth conspiracy theories.
Flat earth conspiracy theories were recommended hundreds of
millions of times. This might sound just funny and, ``Oh, look
at those people,'' but actually this is very serious. I have a
researcher friend who studied this. If the flat Earth theory is
true, it means not just that all of government is lying to you,
but all of science is lying to you. So think about that for a
second. That is like a meltdown of all of our rational
epistemic understanding of the world.
And, as you said, these things are autoplaying. So autoplay
is just like [holds up cup]--it hacks your brain's stopping
cue. So, as a magician, how do I know if I want you to stop? I
put a stopping cue and your mind wakes up. It is like a right
angle in a choice. If I stop drinking, if the water hits the
bottom of the glass, I have to make a conscious choice, do I
want more? But we can design it so the bowl never stops. We can
just keep refilling the water, and you never stop. And that is
how we basically have kept millions of kids addicted. In places
like the Philippines, people watch YouTube for 10 hours a day.
Ten hours a day.
Ms. Castor. This has significant cost to the public, and
that is one of the points I hope people will understand. As Dr.
Donovan says, there is economy of misinformation now. These
online platforms now are passing along--they are monetizing,
making billions of dollars. Meanwhile, public health costs, law
enforcement costs are adding up to the public, and we have a
real responsibility to tackle this and level the playing field.
Mr. Harris. And by not acting, we are subsidizing our
societal self-destruction. I mean, we are subsidizing that
right now. So yes, absolutely. Thank you so much.
Ms. Schakowsky. I recognize Representative Burgess for 5
minutes.
Mr. Burgess. Thank you. Thanks for holding this hearing. I
apologize. We have another Health hearing going on upstairs, so
it is one of those days you got to toggle between important
issues.
Mr. Hurwitz, let me start by asking you--and this is a
little bit off topic, but it is important. In 2018, United
States District Court for Western Pennsylvania indicted seven
Russians for conducting a physical cyber hacking operation in
2016 against Western targets, including the United States Anti-
Doping Agency, in response to the revelation of Russia's state-
sponsored doping campaign. These hackers were representatives
of the Russian military, the GRU. According to the indictment,
the stolen information was publicized by the GRU as part of a
related influence and disinformation campaign designed to
undermine the legitimate interests of the victims. This
information included personal medical information about United
States athletes.
So these GRU hackers used fictitious identities and fake
social media accounts to research and probe victims and their
computer networks. While the methods we are talking about today
are largely in the context of perhaps deceiving voters or
consumers, the harmful potential effects is actually quite
large.
So, in your testimony, you defined the dark pattern, the
practice of using design to prompt desired, if not necessarily
desirable, behavior. Can these dark patterns be used to surveil
people and find ways to hack them in the service of broader
state-sponsored operations?
Mr. Hurwitz. Yes, absolutely, they can. And this goes to
the broader context in which this discussion is happening. We
are not only talking about consumer protection, we are talking
about a fundamental architecture. The nature, as I said before,
of trust online is different. All of those cues that we rely on
for you to know who I am when you see me sitting here. We have
gone through some vetting process to be sitting here. We have
identities. We have telltale cues that you can rely on to know
who I am and who you are. Those are different online, and we
need to think about trust online differently.
One example that I will highlight that goes to an industry-
based solution and, more important, the nature of how we need
to think about these things differently, in the context of
targeted advertising and political advertising in particular,
how do we deal with targeted misinformation for political ads?
Well, one approach which Facebook has been experimenting
with is, instead of saying you can't speak, you can't
advertise, if I target an ad at a group of speakers, Facebook
will let someone else target an ad to that same group, or they
have been experimenting with this.
It is a different way of thinking about how we deal with
establishing trust or responding to untrustworthy information.
We need more creative thinking. We need more research about how
do we establish trust in the online environment.
Mr. Burgess. Well, thank you, and thank you for those
observations.
Ms. Bickert, if I ever doubted the power of Facebook, 3
years ago that doubt was completely eliminated. One of your
representatives actually offered to do a Facebook event in the
district that I represent in northern Texas. And it was not a
political--it was a business-to-business. It is how to
facilitate and run your small business more efficiently. And
wanted to do a program, and we selected a Tuesday morning. And
I asked how big a venue should we get, thinking maybe 20, 30.
And I was told 2,000, expect 2,000 people to show up. I am
like, ``Two thousand people on a Tuesday morning for a
business-to-business Facebook presentation? Are you nuts?''
The place was standing room only, and it was the power of
Facebook getting the word out there that this is what we are
doing. And it was one of the most well-attended events I have
ever been to as an elected representative. So, if I had ever
doubted the power of Facebook, it was certainly brought home to
me just exactly the kind of equity that you are able to wield.
But recognizing that, do you have a sense of the type of
information on your platforms that needs to be fact-checked,
because you do have such an enormous amount of equity?
Ms. Bickert. Yes, Congressman. And thank you for those
words. We are concerned not just with misinformation--that is a
concern, and that is why we developed the relationships we have
now with more than 50 fact-checking organizations--but we are
also concerned with abuse of any type. I am responsible for
managing that, so whether it is terror propaganda, hate speech,
threats of violence, child exploitation content, content that
promotes eating disorders. Any of that violates our policies,
and we go after it proactively to try to find it and remove it.
That is what my team is.
Mr. Burgess. Do you feel you have been successful?
Ms. Bickert. I think we have had a lot of successes, and we
are making huge strides. There is always more to do. We have
begun publishing reports in the past year and a half or so,
every 6 months, where we actually show across different abuse
types how prevalent is this on Facebook from doing a sample,
how much content did we find this quarter and remove, and how
much did we find before anybody reported it to us?
The numbers are trending in a good direction, in terms of
how effective our enforcement measures are, and we hope that
will continue to improve.
Mr. Burgess. As policymakers, can we access that fund of
data to, say, for example, get the number of antivaccine issues
that have been propagated on your platform?
Ms. Bickert. Congressman, I can follow up with you on the
reports we have and any other information.
Mr. Burgess. Thank you. I will yield back.
Ms. Schakowsky. If I could just clarify that question. Is
that information readily available to consumers, or no?
Ms. Bickert. Chairwoman, the reports I just mentioned are
publicly available, and we can follow up with any detailed
requests as well.
Ms. Schakowsky. I recognize Mr. Veasey for 5 minutes for
questioning.
Mr. Veasey. Thank you, Madam Chair. Outside of self-
reporting, what can be done to help educate communities that
may be specifically targeted by, you know, all these different
platforms?
I was wondering, Mr. Harris, if you could address that
specifically, just because I think that a great deal of my
constituency, and even on the Republican side, I think, a great
deal of their constituencies, are probably being targeted,
based on things like race and income, religion, and what have
you.
And is there anything outside of self-reporting that can be
done to just help educate people more?
Mr. Harris. Yes, there are so many things here. And, as you
mentioned, in the 2016 election Russia targeted African-
American populations. I think people don't realize--I think
every time a campaign is discovered, how do we back-notify
people, all of whom were affected, and say ``You were the
target of an influence operation''?
So right now, every single week, we hear reports of Saudi
Arabia, Iran, Israel, China, Russia, all doing various
different influence operations. Russia was recently going after
U.S. veterans. Many veterans would probably say that is a
conspiracy theory, right? But Facebook is the company that
knows exactly who was affected, and they could actually back-
notify every time there is an influence operation, letting
those communities know that this is what happened, and that
they were targeted.
We have to move from ``This is a conspiracy theory'' to
``This is real.'' I have studied cult deprogramming for a
while, and how do you wake people up from a cult when they
don't know they are in? You have to show them essentially the
techniques that were used on them to manipulate them. And every
single time these operations happen, I think that has to be
made visible to people.
And just like we said, you know, we have laws and
protections. We have a Pentagon to protect our physical
borders. We don't have a Pentagon to protect our digital
borders, and so we depend on however many people Facebook
chooses to hire for those teams. One example of this, by the
way, is that the City of Los Angeles spends 25 percent of its
budget on security. Facebook spends 6 percent of its budget on
security, so it is underspending the City of L.A. by about 4
times.
So, you know, you can just make some benchmarks and say,
``Are they solving the problem?'' They have got 2.2 billion
fake accounts, Facebook has, that they took down, fake
accounts. So they have 2.7 billion real accounts, and then
there were 2.2 billion fake accounts. And, you know, I am sure
they got all of them I think would be the line to use here.
Mr. Veasey. Ms. Bickert, you know, given the fact that it
does seem like these foreign agents, these foreign actors, are
targeting people specifically by their race, by their
economics, by what region of the country that they live in, is
Facebook doing anything to gather information or to look at how
specific groups are being targeted?
If African Americans are being targeted for political
misinformation, if whites that live in rural America, if they
are being targeted for political misinformation, if people
based on their likes--like, if you could gatherinformation, if
these foreign actors could gather information based on people
based on things that they like.
So let's say that you were white and you lived in rural
America and you liked One America News and you like these other
things and you may be more likely to believe in these sorts of
conspiracy theories. Are you sure that some of the things that
people are sharing on your platform, the likes and dislikes,
aren't being used as part of that scheme as well?
Could you answer both of those?
Ms. Bickert. Yes, Congressman. Thank you for the question.
There are, broadly speaking, two things that we do. One is
trainings and tools to help people--especially those who might
be most at risk--recognize ways to keep themselves safe from
everything from hacking to scams and other abuse.
Separately, whenever we remove influence operations under
our, what we call this coordinated inauthentic behavior--we
have removed more than 50 such networks in the past year--any
time we do that, we are very public about it, because we want
to expose exactly what we are seeing. And we will even include
examples in our post saying, here is a network, it was in this
country, it was targeting people in this other country, here
are examples of the types of posts that they were putting in
their pages. We think the more we can shine a light on this,
the more we will be able to stop it.
Mr. Veasey. Before my time expires, but if people are being
scientifically--if their likes, and Dr. Burgess' district being
specifically targeted because of certain television or news
programming that they like, if they are African Americans that
are being specifically targeted because Russian actors may
think that they lean a certain way in politics, don't you think
that information ought to be analyzed more closely instead of
relying on--instead of just leaving it up to the user to be
able to figure all of this out? Especially when people work odd
hours and may only have time to digest what they immediately
read, and they may not have an opportunity to go back and
analyze something so deeply as far as what you are saying.
Ms. Bickert. Congressman, I appreciate that. And I will
say, attribution is complicated, and understanding the intent
behind some of these operations is complicated. We think the
best way to do that is to make them public.
And we don't just do this ourselves. We actually work hand-
in-hand with academics and security firms who are studying
these types of things, so that they can see. And sometimes we
will say as we take down a network, ``We have done this in
collaboration or conversation with,'' and we will name the
group.
So there are groups who can look at this and together
hopefully shine light on who the actors are and why they are
doing what they are doing.
Mr. Veasey. Thank you. I yield back.
Ms. Schakowsky. I recognize Mr. Latta for 5 minutes.
Mr. Latta. Well, thank you, Madam Chair, and thanks very
much for holding this very important hearing today. And thank
you to our witnesses for appearing before us. And it is really
important for Americans to get this information.
In 2018, the experts out there estimated that criminals
were successful in stealing over $37 billion from our older
Americans through different scams through the internet,
identity theft, friends, family abuse and impostor schemes. And
last year in my district, I had the Federal Trade Commission
and the IRS out for a senior event, so that the seniors could
be educated on the threat of these scams and how to recognize,
avoid, ward off, and how to recover from them.
Congress recognized that many of these scams were carried
out through the use of manipulative and illegal robocalls. To
combat these scams, I introduced the STOP Robocalls Act, which
was recently signed into law as part of the tray stack, which I
am very glad the President signed over the Christmas holiday.
While I am glad that we were able to get this done, I
continue to be concerned with the ability of scammers to evolve
and adapt to changes in the law by utilizing new technologies
and techniques like deep- and cheapfakes.
And, Ms. Bickert, I don't want to pick on you, and I truly
appreciate you being here today, especially since you are a
little under the weather. And I also appreciated reading your
testimony last night. I found it very interesting and
enlightening.
I have several questions. As more and more seniors are
going online and joining Facebook to keep in contact with their
family, friends, and neighbors, in your testimony, you walk us
through Facebook's efforts to recognize misinformation and what
the company is doing to combat malicious actors using
manipulated media. Is Facebook doing anything specifically to
help protect seniors from being targeted on the platform, or
educating them on how to recognize fake accounts or scams?
Ms. Bickert. Thank you for the question. We are, indeed.
And that includes both in-person trainings for seniors, which
we have done and will continue to do. We also have a guide that
can be more broadly distributed that is publicly available that
is a guide for seniors on the best ways to keep themselves
safe.
But I want to say more broadly, and as somebody who was a
Federal criminal prosecutor for 11 years, looking at that sort
of behavior, this is something we take seriously across the
board. We don't want anybody to be using Facebook to scam
somebody else, and we look proactively for that sort of
behavior and remove it.
Mr. Latta. Just a quick followup. I think it is really
important because, you know, from what we have learned in a lot
of times is that seniors don't want to report things, because
they are afraid that, boy, you know, ``I have been taken. I
don't want to tell my relatives, I don't want to tell my
friends,'' because they are afraid of losing some of what they
might have, and not just on the money side, but how they can
get out there.
And so, I think it is really important that we always think
about our seniors, and just to follow up, because at the
workshop that we had in the District last year, the FTC stated
that one of the best ways to combat scams is to educate the
individuals on how to recognize the illegal behavior so they
can turn that into educating their friends and neighbors.
In addition to your private-sector partnerships, would
Facebook be willing to partner with agencies like the FTC to
make sure the public is informed about scammers operating on
their platform?
Ms. Bickert. Congressman, I am very happy to follow up on
that. We think it is important for people to understand the
tools that are available to keep themselves safe online.
Mr. Latta. Ms. Donovan.
Dr. Donovan. Yes, one of the things that we should also
consider is the way in which people are targeted by age for--I
have looked at reverse mortgage scams, retirement funding
scams, fake healthcare supplements. You know, when you do
retire, it becomes very confusing. You are looking for
information. And if you are looking primarily on Facebook and
then posting about it, you might be retargeted by the
advertising system itself.
And so, even when you are not information-seeking,
Facebook's algorithms and advertising are giving other third
parties information, and then serving advertising to seniors.
And so it is a persistent problem.
Mr. Latta. Thank you. Again, Ms. Bickert, if I can just
follow up quickly with my remaining 30 seconds. Many of the
scammers look for ways to get around Facebook's policies,
including through the development and refinement of new
technologies and techniques.
Is Facebook dedicating the resources and exploring ways to
proactively combat scams instead of reacting after the fact?
Ms. Bickert. Yes, Congressman, we are. I have been
overseeing content policies at Facebook for about 7 years now,
and in that time I would say that we have gone from being
primarily reactive in the way that we enforce our policies to
now primarily proactive. We are really going after abusive
content and trying to find it. We grade ourselves based on how
much we are finding before people report it to us, and we are
now publishing reports to that effect.
Mr. Latta. Thank you very much.
Madam Chair, my time is expired, and I yield back.
Ms. Schakowsky. The gentleman yields back.
And I now recognize Mr. O'Halleran for 5 minutes.
Mr. O'Halleran. I want to thank the chairwoman for holding
this important and timely meeting here today--hearing. I echo
the concerns of my colleagues. The types of deceptive online
practices that have been discussed today are deeply troubling.
I have continually stressed that a top priority for Congress
should be securing our U.S. elections.
We could see dangerous consequences if the right tools are
not in place to prevent the spread of misinformation online.
This is a national security concern. As a former law
enforcement officer, I understand that laws can be meaningless
if they are not enforced. I look forward to hearing more from
our witnesses about the FTC's capabilities and resources to
combat these deceptive online practices.
Dr. Donovan, in your testimony you say that regulatory
guardrails are needed to protect users from being misled
online. I share your concerns about deception and manipulation
online, including the rise in use of the dark patterns,
deepfakes and other kinds of bad practices that can harm
consumers.
Can you explain in more detail what sort of regulatory
guardrails are necessary to prevent these instances?
Dr. Donovan. I will go into one very briefly. One of the
big questions is, if I post something online that is not an
advertisement, you know, I am just trying to inform my known
networks. The problem isn't necessarily always that there is a
piece of fake content out there. The real problem is the scale,
being able to reach millions.
In 2010, 2011, we lauded that as a virtue of platforms. It
really emboldened many of our important social movements and
raised some incredibly important issues. But that wasn't false
information. It wasn't meant to deceive people. It wasn't meant
to siphon money out of other groups. At that time too, you
weren't really able to scale donations. It was much harder to
create networks of fake accounts and pretend to be an entire
constituency.
And so, when I talk about regulatory guardrails, we have to
think about distribution differently than we think about the
content. And then we can also assuage some of the fears that we
have about freedom of expression by looking at what are the
mechanisms by which people can break out of their known
networks? Is it advertising? Is it the use of fake accounts?
How are people going viral? How are posts going viral,
information going viral?
The other thing I would like to know from the government
perspective is, does the FTC have enough insight into platforms
to monitor that, to understand that? And if they don't, if they
don't know why and how tens of millions of dollars are being
siphoned out of Trump's campaign, then that is also another
problem, and we have to think about what does transparency,
what does auditing look like in a very meaningful way.
Mr. O'Halleran. Doctor, do you believe, then, that the FTC
has the adequate authority under Section 5 of the FTC Act to
take action against individuals and companies engaged in
deceptive behavior practices online? And I do want to point out
a Wall Street Journal report that said of the millions of
dollars--200-and-some million dollars--of fines, that they have
only collected about $7,000 since 2015.
Dr. Donovan. Wow. I think that you do have to look a lot
closer at what the FTC has access to and how they can make that
information actionable. For example, proving that there is
substantial injury, if only one group has access to the known
cost or knows the enormity of a scam, then we have to be able
to expedite the transfer of data and the investigation in such
a way that we are not relying on journalists or researchers or
civil society organizations to investigate. I think that the
investigatory powers of the FTC have to also include assessing
substantial injuries.
Mr. O'Halleran. Thank you, Doctor.
Mr. Harris, do you believe the agency has enough resources
to responsibly, swiftly, and appropriately address the issues?
And I just want to point out that we flat-line them all the
time. And on the other side, industry continues to expand at
exponential rates.
Mr. Harris. That is the issue that you are pointing to, is
that the problem-creating aspects of the technology industry,
because they operate at exponential scales, create exponential
issues, harms, problems, scams, et cetera. And so how do you,
you know, have a small body reach such large capacities? This
is why I am thinking about how can we have a digital update for
each of our different agencies who already have jurisdiction
over, whether it is public health or children or scams or
deception, and just have them ask the questions that then are
forced upon the technology companies to use their resources to
calculate, report back, set the goals for what they are going
to do in the next quarter.
Mr. O'Halleran. Thank you, Mr. Harris.
And I yield.
Ms. Schakowsky. The Chair now recognizes Mr. Carter for 5
minutes.
Mr. Carter. Thank you, Madam Chair.
And thank all of you for being here. This is extremely
important, and extremely important to all of our citizens.
I want to start by saying that, you know, when we talk
about deepfake and cheapfake, to me, that is somewhat black and
white. I can understand it. But, Mr. Hurwitz, when we talk
about dark patterns, I think that is more gray in my mind. And
I will just give you an example.
I was a retailer for many years. And I grew up in the
South, OK? We had a grocery store chain, some of you may be
familiar with it: Piggly Wiggly. Now, I always heard that the
way they got their name--and I tried to fact-check it, but I
couldn't find it, but anyway--I always heard the way they got
their name is they arranged their stores to when you went in
you had to kind of wiggle all the way around before you could
get back out so that you would buy more things. It was like a
pig wiggling through the farmyard or something. And they came
up with Piggly Wiggly. Well, that is marketing.
And, you know, another example is all of us go to the
grocery store. When we are at the grocery store and you are in
the checkout line, you got all these things up there that they
are trying to get you to buy. They are not necessarily--you
could argue that they are impulse items. But then again, you
could also make the argument that when you get home you say,
``Geez, I wish I had gotten that at the grocery store. I wish I
would have gotten these batteries or Band-Aids'' or whatever.
How do you differentiate between what is harmful and what
is beneficial?
Mr. Hurwitz. A great question, because it is gray. And, as
I said previously, dark patterns, the term itself is a dark
pattern intended to make us think about this as dark. There are
some clear categories, clear lies, clear false statements,
where we are talking about classic deception. That is pretty
straightforward.
But when we are talking about more behavioral nudges, it
becomes much more difficult. Academics have studied nudges for
decades at this point, and it is hard to predict when they are
going to be effective, when they are not going to be.
In the FTC context, the deception standard has a
materiality requirement. So there needs to be some
demonstration that a practice is material to the consumer harm,
and that is a good sort of framework. If we don't have some
sort of demonstrable harm requirement and causal connection
there--I am a law professor, causation is a basic element of
any legal claim. If you don't have some ability to tie the act
to the harm, you are in dark waters for due process.
Mr. Carter. So do you think we should be instructing the
FTC to conduct research on this as to what is going on here?
Mr. Hurwitz. I think more information is good information.
The FTC is conducting some hearings already. I think greater
investigation is very powerful, both so that the FTC
understands what they should be doing so they can use this
information to establish rules. Where materiality is difficult
to establish, the FTC can issue a rule, go through a rulemaking
process which makes it easier to substantiate an enforcement
action subsequently.
And even to respond, in part, to a previous question, to
the extent that one of the FTC's core powers, even if it
doesn't lack this as an enforcement authority, is to report to
this body and say, ``Look, we are seeing this practice. It is
problematic. We don't have the authority. Can you do something
about it?'' And perhaps this body will act and give it power,
perhaps this body will take direct action, or perhaps the
platforms and other entities will say, ``Oh, wow, the jig's up,
we should change our practices before Congress does something
that could be even more detrimental to us.''
Mr. Carter. Right. Mr. Harris, did you have something?
Mr. Harris. Yes. I have studied this topic for also about a
decade. So you asked what is different about this. You have got
the pig going through the thing. You have got the supermarket
aisle. You have got the last-minute of, sort of last-minute-
purchase items. There are two distinct things that are
different.
The first is that this is infrastructure we live by. When
you talk about children waking up in the morning and you have
autoplay, that is not like the supermarket where I occasionally
go there and I just made some purchases and I am at the very
end of it, and that is the one moment, the one little
microsituation of deception or marketing, which is OK.
In this case, we have children who are, like, spending 10
hours a day. So imagine a supermarket, you are spending 10
hours a day, and you wake up in that supermarket. And so that
is the degree of intimacy and sort of scope in our lives. That
is the first thing.
The second thing is the degree of asymmetry between the
persuader and the persuadee. So, in this case, you have got
someone who knows a little bit more about marketing who is
arranging the shelf space so that the things in the top are at
eye level versus at bottom level. That is one very small amount
of asymmetry.
But in the case of technology, we have a supercomputer
pointed at your brain, meaning like the Facebook news feed
sitting there, and using the vast resources of 2.7 billion
people's behavior to calculate the perfect thing to show you
next and to not be discriminant about whether it is good for
you, whether it is true, whether it is trustworthy, whether it
is credible. And so, it knows more about your weaknesses than
you know about yourself, and the degree of asymmetry is far
beyond anything we have experienced.
Mr. Carter. And you want the Federal Government to control
that?
Mr. Harris. I think we have to ask questions about--when
there is that degree of asymmetry, about intimate aspects of
your weaknesses, and its business model is to exploit that
asymmetry. It is as if a psychotherapist who knows everything
about your weaknesses uses it with a for-profit advertising
business model.
Mr. Hurwitz. The challenge is that can also go the other
way. It can used to strengthen.
Mr. Carter. Yes, yes.
Mr. Hurwitz. Mr. Harris used the example earlier of what if
autoplay is shifting us towards conspiracy theories. OK, that
is a dark pattern, that is bad. What if, instead, it was using
us to shift us the other way, to the light, to greater
education. If we say autoplay is bad, then we are taking both
of those options off the table.
This can be used for good, and the question that you asked
about how do we differentiate between good uses and bad, that
is the question.
Mr. Carter. Thank you, Madam Chair. I yield back.
Ms. Schakowsky. Mr. Cardenas is recognized for 5 minutes.
Mr. Cardenas. Thank you, Madam Chair, and thank you so much
for holding this very important hearing that, unfortunately, I
think most Americans don't understand how important this is to
every single one of us, especially to our children and future
generations.
There is an app, TikTok, question mark. Is it a deepfake
maker? Five days ago, TechCrunch reported that ByteDance, the
parent company of the popular video-sharing app TikTok, may
have secretly built a deepfake maker. Although there is no
indication that TikTok intends to actually introduce this
feature, the prospect of deepfake technology being made
available on such a massive scale and on a platform that is so
popular with kids raises a number of troubling questions.
So my question to you, Mr. Harris, is in your testimony you
discuss at length the multitude of ways that children are
harmed by new technology. Can you talk about why this news may
be concerning?
Mr. Harris. Yes. Thank you for the question.
So deepfakes is a really complex issue. I think if you look
at how other governments are responding to this--I don't mean
to look at China for legal guidance, but they see this as so
threatening to their society, the fabric of truth and trust in
their society, that if you post a deepfake without labeling it
clearly as a deepfake, you can actually go to jail.
So they are not saying if you post a deepfake you go to
jail. They are saying if you post it without labeling it, you
go to jail. You can imagine a world where Facebook says, ``If
you post a deepfake without labeling it, we actually maybe
suspend your account for 24 hours, so that you sort of feel--
and we label your account to other people who see your
account----''
Mr. Cardenas. Hold on a second. My colleague on the other
side of the aisle just warned, quote, ``And you want to have
the government control this?'' You just gave an example of
where private industry could, in fact, create deterrents----
Mr. Harris. That is right.
Mr. Cardenas [continuing]. To bad behavior, not the
government, but actual industry. OK, go ahead.
Mr. Harris. So that is right. And so they can create--and
that is the point, is instead of using these AI Whac-a-Mole
approaches where the engineers at Facebook--how many engineers
at Facebook speak the 22 languages of India where there was an
election last year? They are controlling the information
infrastructure not just for this country, but for every
country, and they don't speak the languages of the countries
that they operate in, and they are automating that.
And, instead of trying to use AI where they are just
missing everything going by--yes, they have made many
investments, we should celebrate that, there are people working
very hard, it is much better than it was before--but they have
created a digital Frankenstein where there is far more content,
advertising, variations of texts, lies, et cetera, than they
have the capacity to deal with.
And so you can't create problems way beyond the scope of
your ability to address them. It would be like creating nuclear
power plants everywhere with the risk of meltdown, without
actually having a plan for security.
Mr. Cardenas. Now, getting back to your example where
industry could, in fact, for example, Facebook could say ``We
are going to suspend your account for 24 hours'' or something
like that, with all due respect, in that example, Facebook
might lose a little bit of revenue, as well as the person that
they are trying to deter from bad action is likely going to
lose revenue as well, correct?
Mr. Harris. That is correct. But maybe that is an
acceptable cost, given we are talking about the total meltdown
of trust.
Mr. Cardenas. Yes, but maybe it is acceptable when you look
at it intellectually and honestly, but when you look at it from
whether or not private industry is going to take it upon
themselves to actually impact their shareholders' revenue, that
is where government has a place and space to get involved and
say, proper actions and reactions need to be put in place so
that people can understand that you can't and you shouldn't
just look at this from a profit center motive.
Mr. Harris. That is right.
Mr. Cardenas. Because in this world sometimes the negative
actions are more profitable for somebody out there than
positive, good actions. And that is one of the things that is
unfortunate.
And you talk about languages around the world, but the
number one target, in my opinion, for these bad actions for
both financial gain and also the tearing down of the fabric of
the democracy of the greatest nation on the planet, the United
States, is the United States, we are the biggest target for
various reasons.
Two main reasons are because we are supposed to be the
shining light on the hill for the rest of the world for what a
good democracy should be like. And secondly, we are by far and
away the largest economy, the biggest consumer group of folks
on the planet.
So, therefore, there is a motive for people to focus on
profit and focus on their negative, bad intentions against our
interests, the interests of the American people. Is that
accurate?
Mr. Harris. That is exactly right. And this is a national
security--I see this as a long-term--I mean, the polarization
dynamics are accelerating towards civil war-level things,
hashtag civilwariscoming.
Our colleague Renee DiResta says, ``If you can make it
trend, you can make it true.'' When you are planting these
suggestions and getting people to even think those thoughts
because you can manipulate the architecture, we are profiting,
as I said, we are subsidizing our own self-destruction if the
government doesn't say that these things can't just be
profitable.
Mr. Cardenas. Thank you to the witnesses. And thank you,
Mr. Harris. I have run out of time. I wish I had more time.
Thank you.
Ms. Schakowsky. The gentleman yields back.
And now I recognize Mr. Soto for 5 minutes.
Mr. Soto. Thank you, Madam Chair.
It has been my experience that a lie seems to be able to
travel faster on the internet than the speed of light, while
the truth always goes at such a snail's pace. I suppose that is
because of the algorithms we see.
I want to start with deepfakes and cheap fakes. We know
through New York Times v. Sullivan that defamation of public
figures requires actual malice. And some of these just appear
to be malicious on their face.
I appreciate the labeling, Ms. Bickert, that Facebook is
doing now. That is something that we actually were pondering in
our office as well. But why wouldn't Facebook simply just take
down the fake Pelosi video?
Ms. Bickert. Thank you for the question.
Our approach is to give people more information so that, if
something is going to be in the public discourse, they will
know how to assess it, how to contextualize it. That is why we
work with the fact checkers.
I will say that in the past 6 months it is feedback from
academics and civil society groups that has led us to come up
with stronger warning screens.
Mr. Soto. Would that be labeled under your current policy
now as false, that video?
Ms. Bickert. I am sorry, which video?
Mr. Soto. Would the fake Pelosi video be labeled as false
under your new policy?
Ms. Bickert. Yes. And it was labeled false. At the time we
did--we think we could have gotten that to fact checkers
faster, and we think the label that we put on it could have
been more clear. We now have the label for something that has
been rated false. You have to click through it so it actually
obscures the image. And it says ``false information.'' And it
says ``This has been rated false by fact checkers.'' You have
to click through it, and you see information from the fact-
checking source.
Mr. Soto. Thanks.
In 2016 there was a fake Trump rally put together by
Russians in Florida, complete with a Hillary Clinton in a
prison and a fake Bill Clinton.
Could a fake rally be created today through Facebook in the
United States by the Russians under existing technology?
Ms. Bickert. The network that created that was fake and
inauthentic, and we removed it. We were slow to find it.
I think our enforcement has gotten a lot better. And, as a
data point for that, in 2016 we removed one such network. This
past year, we removed more than 50 networks. Now, that is a
global number all over the world. But these are organizations
that are using networks of accounts--some fake, some real--in
an attempt to obscure who they are or to push false
information.
Mr. Soto. So could it happen again right now?
Ms. Bickert. Our enforcement is not perfect. However, we
have made huge strides, and that is shown by the dramatic
increase in the number of networks that we have removed.
And I will say that we do it not just by ourselves, but we
work with security firms and academics who are studying this to
make sure we are staying on top of it.
Mr. Soto. What do you think Facebook's duty is, as well as
other social media platforms, to prevent the spread of lies
across the internet?
Ms. Bickert. I am sorry. Could you repeat that?
Mr. Soto. What you do think Facebook and other social
platforms' duty is to prevent the spread of lies across the
internet?
Ms. Bickert. I can speak for Facebook. We think it is
important for people to be able to connect safely and with
authentic information. And my team is responsible for both.
So there is our approach to misinformation where we try to
get people--label contented as false and get them accurate
information. And then there is everything we also do to remove
abusive content that violates our standards.
Mr. Soto. Thank you, Ms. Bickert.
Dr. Donovan, I saw you reacting to the fake Trump rally
aspect. Could that still happen now under existing safeguards
in social media?
Dr. Donovan. Yes. And the reason why it can still happen is
because the platform's openness is now turning into a bit of a
vulnerability for the rest of society.
So what is dangerous about events like that is the kind of
research we do, we are often trying to understand, well, what
is happening online? And what happens when the wires--the
interaction between the wires and the weed? Like when people
start to be mobilized, start to show up places, that to us is
one order of magnitude much more dangerous.
Mr. Soto. What do you think we should be doing as
government to help prevent something like that?
Dr. Donovan. There are ways in which I think, when people
are using particularly events features, group features, there
has to be added transparency about who, what, when, where those
events are being organized by.
And there have been instances in Facebook very recently
where they have added transparency pages, but it is not always
clear to the user who is behind what page and for what reason
they are launching a protest.
What is dangerous, though, is that actual constituents show
up, real people show up as fodder for this. And so we have to
be really careful that they don't stage different parties like
they did in Texas across the street from one another at the
same time. And so we don't want to have manipulation that
creates this serious problem for law enforcement, as well as
others in the area.
Mr. Soto. Thanks. My time has expired.
Ms. Schakowsky. I now recognize Congresswoman Matsui for 5
minutes.
Ms. Matsui. Thank you very much, Madam Chair. And I really
appreciate the witnesses here today, especially on this really
important issue.
I introduced the Blockchain Promotion Act with Congressman
Guthrie to direct the Department of Commerce to convene a
working group of stakeholders to develop a consensus-based
definition of blockchain. Currently there is no common
definition, which has hindered its deployment.
Blockchain technology could have interesting applications
in the communication space, including new ways of identity
verification. This technology is unique in that it can help
distinguish between credible and noncredible news sources in a
decentralized fashion, rather than relying on one company or
organization to serve as a sole gatekeeper.
I have a lot of questions. I would like succinct answers to
this.
Ms. Donovan, do you see value in promoting impartial,
decentralized methods of identity verification as a tool to
combat the spread of misinformation?
Dr. Donovan. I think in limited cases, yes, especially
around purchasing of advertising, which is allowing you to
break out of your known networks and to reach other people,
especially if those advertising features do allow you to target
very specific groups.
I am interested in learning more about this consensus on
definition, because I also think it might help us understand
what is a social media company, what are their--how do we
define their broadcast mechanisms, how do we define them
related to the media, media company, as well as the other kinds
of products that they build. And I think it would also get us a
lot further in understanding what it is we say when we say
deepfakes or even AI.
Ms. Matsui. OK. The European Commission has recently
announced that it will be supporting research to advance
blockchain technology to support a more accurate online news
environment.
The entire panel, just a yes or no is sufficient.
Do you believe the U.S. should be keeping pace with Europe
in this space? Yes or no?
As far as blockchain, do you think that the European
Commission is supporting research to advance blockchain
technology to support a more accurate online news development?
Do you believe that the U.S. should be keeping pace with Europe
regarding this?
Ms. Bickert. This is not my area.
Ms. Matsui. OK. Dr. Donovan, I probably would say----
Dr. Donovan. Yes, more research could help us understand
this better.
Ms. Matsui. Mr. Hurwitz, yes or no?
Mr. Hurwitz. Around the world, many are outpacing us in
blockchain.
Ms. Matsui. OK.
Mr. Harris?
Mr. Harris. It is not my area, but I know that China is
working on a decentralized currency and could basically get all
of the countries in which it is indebting them to their
infrastructure with these huge Belt and Road plans. If they
switch the global currency to their decentralized currency,
that is a major national security threat and would change the
entire world order. I think much more work has to be done in
the U.S. to protect against China gaining currency advantage
and changing the world of reserve currency.
Ms. Matsui. Thank you.
It is an undisputed fact, reaffirmed by America's
intelligence agencies, that Russia interfered in our 2016 and
2018 elections through targeted and prolonged online campaigns.
We know that Russia is ramping up for 2020, and the American
voters will once again be exposed to new lies, falsehoods, and
misinformation designed to sow division in our democratic
process.
While I was glad to see the recent funding bill included
$425 million in election security grants, this is only part of
a much larger solution. To protect the most fundamental
function of our democracy, social media companies need to take
clear, forceful action against foreign attempts to interfere
with our elections.
Mr. Harris, how have the various election interference
strategies evolved from the 2016 and 2018 election cycles?
Mr. Harris. You know, I am actually not an expert on
exactly what Russia is doing now. What I will say is I think
that we need a mass public awareness campaign to inoculate the
public. Think of it as like a cultural vaccine.
And there is actually precedent in the United States for
this. So, back in the 1940s, we had the Committee for National
Morale and the Institute for Propaganda Analysis that actually
did a domestic awareness campaign about the threat of fascist
propaganda.
You have probably seen the videos from--they are black and
white--from 1947. It was called ``Don't Be a Sucker.'' And they
had us looking at a guy spouting fascist propaganda, someone
starting to nod, and then the guy taps him on the shoulder and
says, ``Now, son, that is fascist propaganda, and here is how
to spot it.''
We actually saw this as a deep threat, a national security
threat to our country. We could have another mass public
awareness campaign now, and we could have the help of the
technology companies to collectively use their distribution to
distribute that inoculation campaign so everybody actually knew
the threat of the problem.
Ms. Matsui. Does the rest of the panel agree with Mr.
Harris on this, to have this public awareness campaign?
Mr. Hurwitz. Probably. I will just note that it runs the
risk of being called a dark pattern if the platforms are
starting to label certain content in certain ways. So there is
a crosscurrent for our discussion to note there.
Ms. Matsui. OK. Well, we don't come to any solutions now,
but I appreciate it. And I have run out of time. Thank you very
much.
Ms. Bickert. Congresswoman, I would just point to the ads
library that we have put in place over the past few years,
which has really brought an unprecedented level of openness to
political advertising. So people can now see who is behind an
ad, who paid for it, and we verify the identity of those
advertisers.
Ms. Matsui. I think it is difficult for most people out
there to really do that, unless it is right in front of them.
But I am glad that that is happening. But I think we should
have much more exposure about this.
Thank you.
Ms. Schakowsky. I now recognize Mr. McNerney for 5 minutes.
Mr. McNerney. I thank the chair.
And I thank the witnesses. Your testimony has been helpful,
and I appreciate it. But I have to say, with big power comes
big responsibility, and I am disappointed, in my opinion, that
Facebook hasn't really stepped up to that responsibility.
Back in June, I sent a letter to Mr. Zuckerberg, and I was
joined by nearly all the Democrats on the committee. In this
letter we noted that we are concerned about the potential
conflict of interest between Facebook's bottom line and
addressing misinformation on its platform. Six months later, I
remain very concerned that Facebook is putting its bottom line
ahead of addressing misinformation.
Ms. Bickert, Facebook's content monetization policy states
that content that depicts or discusses subjects in the
following categories may face reduced or restricted
monetization, and misinformation is included on the list. It is
troubling that your policy doesn't simply ban misinformation.
Do you think there are cases where misinformation can and
should be monetized? Please answer yes or no.
Ms. Bickert. Congressman, no. If we see somebody that is
intentionally sharing misinformation, and we make this clear in
our policies, they will lose the ability to monetize.
Mr. McNerney. OK. Well, that sounds different than what is
in your company's stated policy.
But the response I received from Facebook to my letter
failed to answer many of my questions. For example, I asked the
following question that was left unanswered, and I would like
to give you a chance to answer it today. How many project
managers does Facebook employ whose full-time job it is to
address misinformation?
Ms. Bickert. Congressman, I don't have a number of PMs. I
can tell you that across my team, our engineering teams, and
our content review teams, this is something that is a priority.
Building that network of the relationships with more than 50
fact-checking organizations is something that has taken the
efforts of a number of teams across the company.
Mr. McNerney. Does that include software engineers?
Ms. Bickert. It does, because there for any of these
programs you need to have an infrastructure that can help
recognize when something might be misinformation, allow people
to report when something might be misinformation, get things
over to the fact-checking organization.
Mr. McNerney. OK. So I am going to ask you to provide that
information, how many full-time employees, including software
engineers who were employed in that, to identify
misinformation.
Ms. Bickert. We are happy to try to follow up and answer.
Mr. McNerney. Another question that was left unanswered is,
on average, from the time a content is posted on Facebook's
platform, how long does it take for Facebook to flag suspicious
content to third-party fact checkers, third-party fact checkers
to review the content, and Facebook to take remedial action
once the content--once the review is completed?
Ms. Bickert. Congressman, the answer depends. This could
happen very quickly. We actually allow fact-checking
organizations to proactively rate content they see on Facebook.
So they----
Mr. McNerney. You think that would be fast enough to keep
deepfakes from going viral or other misinformation from going
viral?
Ms. Bickert. If they rate something proactively then it
happens instantly. And we also use technology and use the
reporting to flag content to them, and we often see that they
will rate it very quickly.
Mr. McNerney. Well, moving on, I am very concerned that
Facebook is not prepared to address misinformation on its
platform in advance of this year's election. Will you commit to
having a third-party audit conducted by June 1 of Facebook's
practices for combating the spread of disinformation on its
platform and for the results of this audit to be made available
to the public?
Ms. Bickert. Congressman, we are very happy to answer any
questions about how we do what we do. We think transparency is
important. And we are happy to follow up with any suggestions
that you have.
Mr. McNerney. I would request a third-party audit--I am not
talking about the civil rights audit--an independent third-
party audit be conducted at Facebook by June 1.
Ms. Bickert. Congressman, again, we are very transparent
about what our policies and practices are, and we are happy to
follow up with any specific suggestions.
Mr. McNerney. Mr. Harris.
Mr. Harris. I was going to say, their third-party fact-
checking services are massively understaffed, underfunded, and
a lot of the people are dropping out of the program. And the
amount of information flowing through that channel is far
beyond their capacity to respond.
More or less, fact checking isn't even really the relevant
issue. I think if you look at the clearest evidence of this, is
Facebook's own employees wrote a letter to Mark Zuckerberg
saying, ``You are undermining our election integrity efforts
with your current political ads policy.''
That says it all to me. That letter was leaked to The New
York Times about a month ago, I think that those people,
because they are closest to the problem, they do the research
queries, they understand how bad the issue is.
We are on the outside. We don't actually know. It is almost
like they are Exxon, but they also own the satellites that
would show us how much pollution there is. So we don't actually
know on the outside. So all we can do is trust people like that
on the inside that are saying this is far less than what we
would like to do. And they still have not updated their policy.
Mr. McNerney. Thank you. I yield back.
Ms. Schakowsky. I recognize Congresswoman Dingell for 5
minutes for questions.
Mrs. Dingell. Thank you, Madam Chair.
And thank you all of you for being here today. This is a
subject that really matters to me, like it does to all of us.
But in the past we have treated what little protections people
have online as something that is separate from those we have in
our day-to-day lives offline. But the line between what happens
online and offline is virtually nonexistent. Gone are the days
when we can separate one from the other.
Millions of Americans have been affected by data breaches
and privacy abuses. The numbers are so large that you can't
even wrap your head around them. I mean, I have talked to
Members here and they don't even at times understand what has
happened or how people have collected data about us.
The resources to help folks protect themselves after the
fact are desperately needed. But what is really happening is
that the cost of failure to protect sensitive information is
being pushed on millions of people who are being breached and
not trying to do anything. It is a market externality.
And that is where the government, I believe, must step in.
You go to the pharmacy to fill a prescription, you assume that
the medicine you are going to get is going to be safe, it is
not going to kill you. If you go outside, you assume that the
air you breathe--you assume--is going to be safe, or we are
trying to make it that way.
And that is because we have laws that protect people from
have a long list of known market externalities and the burden
isn't placed on their ability to find out is the medicine you
are taking OK, safe, and is the air you are breathing clean. We
are still working on that, but it is one we have identified. It
shouldn't be any different for market externalities that are
digital.
Ms. Bickert, I will admit I have sent a letter to Facebook
today which has a lot of questions that didn't lend themselves
to answer here, so I hope that they will be answered.
But I would like to get yes-or-no answers from the panel on
the following questions. And I am going start this way, with
Mr. Harris, because we always start with you, Ms. Bickert, and
we will give you a little--and thank you for being here even
though you are sick.
Do you believe that the selling of real-time cell phone
location without users' consent constitutes a market
externality?
Mr. Harris?
Mr. Harris. I don't know with that specific one, but the
entire surveillance capitalism system produces vast harms that
are all on the balance sheets of societies, whether that is the
mental health of children, the manipulation of elections, the
breakdown of polarization.
Mrs. Dingell. But it is a market externality.
Mr. Harris. Absolutely, all market externality.
Mrs. Dingell. OK, let's go down.
Mr. Hurwitz?
Mr. Hurwitz. Based on the economic definition of an
externality, no, it is not. However, it can be problematic.
Mrs. Dingell. Dr. Donovan?
Dr. Donovan. I am in line with Gus.
Mrs. Dingell. Ms. Bickert?
Ms. Bickert. I am not an economist, but we do think user
consent is very important.
Mrs. Dingell. Second question: Yes or no, do you believe
that having 400 million pieces of personally identifiable
information made public, including passport numbers, names,
addresses, and payment information, is a market externality?
Mr. Harris?
Mr. Harris. Similarly, on sort of classic economic
definition, I don't know if that would specifically qualify,
but it is deeply alarming.
Mr. Hurwitz. Same answer.
Dr. Donovan. Agreed.
Ms. Bickert. Same answer.
Mrs. Dingell. So are you all agreeing with Mr. Harris?
Mr. Hurwitz. Same answer as I gave previously. It is not
the technical economic definition.
Mrs. Dingell. I just wanted to see if we had gotten you to
understand what a bother it is.
Three, do you believe that having 148 million individuals'
personally identifiable information, including credit card
numbers, driver's license, and Social Security numbers, made
public is a market externality?
Mr. Harris?
Mr. Harris. I can see it is sort of like an oil spill
externality.
Mrs. Dingell. Mr. Hurwitz?
Mr. Hurwitz. The same answer.
Mrs. Dingell. So you don't think it is a problem.
Mr. Hurwitz. I don't--I don't not think it is a problem. I
wouldn't characterize it as an externality and use it as a----
Mrs. Dingell. Do you not think we have got to protect
people from that?
Mr. Hurwitz. No, that is not what I am saying. I have an
economics background. I rely on a more technical definition of
an externality.
Mrs. Dingell. Dr. Donovan?
Dr. Donovan. It is an incredibly important problem.
Mrs. Dingell. Ms. Bickert?
Ms. Bickert. Yes, I would echo Dr. Donovan.
Mrs. Dingell. Do you believe that having the data of 87
million users taken and used for nefarious and political
purposes is a market externality?
Mr. Harris?
Mr. Harris. I think it is the same answer as before.
Mr. Hurwitz. If I break into your house and steal your
stuff and sell it on the black market, that is not an
externality. However, it is a problem.
Mrs. Dingell. Dr. Donovan?
Dr. Donovan. Well, I wouldn't characterize it as a break-
in. It was facilitated by the features built into the platform,
and it is a huge problem.
Mrs. Dingell. Thank you.
Ms. Bickert?
Ms. Bickert. Again, we think that user control and consent
is very important.
Mrs. Dingell. Last question. I am out of time, so you are
going to have to be fast.
And finally, do you believe that simply asking whoever took
it to please delete it is an appropriate response?
Mr. Harris?
Mr. Harris. It is very hard to enforce that. And once the
data is out there, it is distributed everywhere. So we have to
live in a world where now we assume that this is just out
there.
Mr. Hurwitz. You need to solve the problem on the front
end.
Mrs. Dingell. Dr. Donovan?
Dr. Donovan. That never should have been allowed in the
first place.
Mrs. Dingell. Ms. Bickert?
Ms. Bickert. Again, we think that it is very important to
give people control over their data, and we are doing our best
to make sure that we are doing that.
Mrs. Dingell. So I am out of time. Thank you, Madam Chair.
Ms. Blunt Rochester [presiding]. Thank you. The gentlewoman
yields. And I recognize myself for 5 minutes.
Thank you to the chairwoman in her absence, and thank you
to the panelists.
This is a vitally important conversation that we are
having. What I have noticed is that technology is outpacing
policy and the people. And so we are feeling the impacts in our
mental health, we are feeling it in our economy, we are feeling
it in our form of government. And so this is a very important
conversation.
And I would like to start with a few questions that are
kind of off of the dark patterns and those issues but really do
deal with the idea of deceptive and manipulative practice. And
it is just a basic question, so yes or no, and it is really
surrounding the platforms that we have and the ability for
people with disabilities to use them.
Are each of you, or any of you, familiar with the term
universal design? And I will just ask Mr. Harris.
Mr. Harris. Vaguely, yes.
Ms. Blunt Rochester. Mr. Hurwitz?
Mr. Hurwitz. Vaguely, yes.
Dr. Donovan. Yes.
Ms. Blunt Rochester. Yes.
Ms. Bickert. Vaguely, yes.
Ms. Blunt Rochester. Vaguely. OK. So there are a lot of
vaguelies, and I don't have time to really talk about what
universal design is. But I think, as we look at how people are
treated in our society, universal design and looking at people
with disabilities is one of the areas that I would like to
follow up with each of you on.
I would now like to turn my time to a discussion about dark
patterns. And every single Member of Congress and every one of
our constituents, virtually everyone, has been affected by this
in some respect. Every day, whether it is giving up our
location data, or manipulated into purchasing products that
they don't need, or providing sensitive information that
enables scams, many of us are targeted.
And, while the failure to address dark patterns harms
individuals, one of the areas that is of deeper concern to me
is the challenge for us as a society as a whole. Cambridge
Analytica, that scandal in and of itself was a great example
for all of us of it wasn't just an individual that was harmed,
it was our society, and we see some of the remnants of it to
this day.
And so I heard someone say to me yesterday that they hoped
that this hearing was not just a hearing, but a real wakeup
call, a wakeup call to our country. And so my first question is
to Mr. Harris.
Do you believe that oversight of dark patterns and the
other deceptive and manipulative practices discussed here are
well suited for industry self-regulation?
Mr. Harris. No, absolutely not.
Ms. Blunt Rochester. And I would like to follow up with Ms.
Bickert.
Does Facebook have a responsibility to develop user
interfaces that are transparent and fair to its users?
Ms. Bickert. We definitely want that. And, yes, I think we
are working on new ways to be transparent all the time.
Ms. Blunt Rochester. Does Section 230 of the Communications
Decency Act provide immunities to Facebook over these issues?
Ms. Bickert. Section 230 is an important part of my team
being able to do what we do. So, yes, it gives us the ability
to proactively look for abuse and remove it.
Ms. Blunt Rochester. But does it provide immunities? You
would say yes?
Ms. Bickert. I am sorry, what is the specific--Section 230
does provide us certain protections. The most important from my
standpoint is the ability for us to go after abuse on our
platform. But separately it is also an important mechanism for
people who use the internet to be able to post to platforms
like Facebook.
Ms. Blunt Rochester. I guess one of my concerns here for
asking that question is we are having a big conversation about
the balance of freedom of speech, in addition to the ability
for people to yell fire in a crowded place. And so I am going
to turn back to Mr. Harris.
How do you think that we in Congress can develop a more
agile and responsive response to the concerning trends on the
internet? You mentioned a digital update of Federal agencies.
Can you talk a little bit about that as well?
Mr. Harris. Just as you said, that the problem here is we
have--this is E.O. Wilson--the problem of humanity is we have
paleolithic emotions, medieval institutions, and accelerating.
godlike technology. When your steering wheel goes about a light
year behind your accelerating, godlike technology, the system
crashes.
So the whole point is, we have to give a digital update to
some of the existing institutions--Health and Human Services,
FCC, FTC, you can imagine every category of society--and saying
where do we already have jurisdiction about each of these
areas, and ask them to come up with a plan for what their
digital update is going to be and put the tech companies in a
direct relationship where every quarter there is an audit and
there is a set of actions that are going to be taken to
ameliorate these harms.
That is the only way I can see scaling this, absent
creating a whole new digital Federal agency, which will be way
too late for these issues.
Ms. Blunt Rochester. I know I am running out of time, but
my other question really was going to be to Ms. Bickert on the
role that you see of government. I think we are having a lot of
conversations here about freedom of speech and also the role of
government.
And so as a followup, I would like to have a conversation
with you about what you see as that role of government versus
self-regulation and how we can make something happen here. The
bigger concern is for us to make sure that we are looking at
this both as an individual level, but also as a society.
And I yield my time and recognize the gentlewoman from New
York, Ms. Clarke.
Ms. Clarke. Thank you very much, Madam Chair.
And I thank our ranking member, I thank our panelists for
their expert witness here today.
Deepfakes currently pose a significant and an unprecedented
threat. Now more than ever, we need to prepare for the
possibility that foreign adversaries will use deepfakes to
spread disinformation and interfere in our election, which is
why I have successfully secured language in the NDAA requiring
notification be given to Congress if Russia or China seek to do
exactly this.
But deepfakes have been and will be used to harm individual
Americans. We have already seen instances of women's images
being superimposed on fake pornographic videos. As these tools
become more affordable and accessible, we can expect deepfakes
to be used to influence financial markets, discredit
dissidents, and even incite violence.
That is why I have introduced the first House bill to
address this threat, the DEEPFAKES Accountability Act, which
requires creators to label deepfakes as altered content,
updates our identity theft statutes for digital impersonation,
and requires cooperation between the government and private
sector to develop detection technologies. I am now working on a
second bill specifically to address how online platforms deal
with deepfake content.
So, Dr. Donovan, cheap fakes. We have often talked about
deepfakes, where the technology footprint of the content has
changed. But can you talk a bit more about the national
security implications of cheap fakes, such as the Pelosi video,
where footage is simply altered instead of entirely fabricated?
Dr. Donovan. One of the most effective political uses of a
cheap fake is to draw attention and shift the entire media
narrative towards a false claim. And so particularly what we
saw last week with the Biden video was concerning because you
have hundreds of newsrooms kick into gear to dispute something,
a video, and platforms have allowed it to scale to a level
where the public is curious and are looking for that content,
and then are also coming into contact with other nefarious
actors and networks.
Ms. Clarke. What would you say can be done by government to
counteract the threat?
Dr. Donovan. There has to be--I think you are moving very
much in the direction I would go to, where we need to have some
labels, we need to understand the identity threat that it
poses, and that there needs to be broader cooperation between
governments.
As well I think that the cost to journalism is very high,
because all of the energy and resources that go into tracking,
mapping, and getting public information out there, I think the
platform companies can do a much better job of preventing that
harm up front by looking at content when it does seem to go
wildly out of scale with the usual activity of an account and
to proactively look at things where, if you do see an uptick of
500,000 views on something, maybe there needs to be proactive
content moderation.
Ms. Clarke. Very well.
Ms. Bickert, Facebook is a founding member of the Deepfake
Technology Challenge, but detection is only partially a
technology issue. We also need to have a definition of what
fake is and a policy for which kind of fake videos are actually
acceptable.
Last summer you informed Congress that Facebook is working
on a precise definition for what constitutes a deepfake. Can
you update us on those efforts, especially in light of your
announcement yesterday? And specifically how do you intend to
differentiate between legitimate deepfakes, such as those
created by Hollywood for entertainment, and malicious ones?
Ms. Bickert. Thank you for the question.
The policy that we put out yesterday is designed to address
the most sophisticated types of manipulated media, and this
fits within the definition of what many academics would call
deepfakes, so that we can remove it.
Now, beyond that, we do think it is useful to work with
others in industry and civil society and academia to actually
have common definitions so we are all talking about the same
thing. And those are conversations that we have been a part of
in the past 6 months. We will continue to be a part of those.
And we are hoping that, working together with industry and
other stakeholders, we will be able to come up with
comprehensive definitions.
Ms. Clarke. Should the intent of the deepfake or rather its
subject matter be the focus?
Ms. Bickert. I am sorry. Could you repeat that?
Ms. Clarke. Should the intent of the deepfake or the
subject matter be the focus?
Ms. Bickert. From our standpoint, it is often difficult to
tell intent when we are talking about many different types of
abuse, but also specifically with deepfakes for misinformation,
and that is why if you look at our policy definition it doesn't
focus on intent so much as what the effects would be on the
viewer.
Ms. Clarke. Thank you very much. I yield back.
I thank you, Madam Chair, for allowing my participation
today.
Ms. Schakowsky [presiding]. That concludes the questioning.
I have things I want to put into the record, and maybe the
ranking member does as well. But I did want to make an ending
comment, and I would welcome her to do the same if she wishes.
So we had a discussion that took us to the grocery store,
but we are now in a new world that we are discussing that is
hugely bigger when we talk about Facebook. And as you say in
your testimony, Facebook is a community of more than 2 billion
people spanning countries, cultures, and languages across the
globe.
But I think that there is now such an incredible and
justified distrust of how we are being protected. We know in
the physical world we do have laws that apply and that
expectations of consumers are that those will be somehow there
to protect us. But in fact they aren't.
We live, then, in the virtual world and the digital world
in a place of self-regulation. And it seems to me that that has
not satisfied expectations of consumers correctly. And we don't
have institutions right now, even when they have the
authorities, have the funding, have the expertise--I am
thinking of the Federal Trade Commission, just as an example--
to do what it needs to do.
But we don't have a regulatory framework at all that I
think, hopefully in a bipartisan way, we can think about. And
it may include things like just the kinds of audits that you
were talking about, Mr. Harris, which would not necessarily
create new regulatory laws, but we may need to.
And to me, that is the big takeaway today. When you have
communities that are bigger than any country in the entire
world that are essentially making decisions for all of the rest
of us, and we know that we have been victimized, that the
Government of the United States of America does need to
respond. That is my takeaway from this hearing.
And I would appreciate hearing from the ranking member.
Mrs. Rodgers. I thank the chair, and I thank everyone for
being here. I think it is important that we all become more
educated.
I wanted to bring to everyone's attention that the FTC is
holding a hearing on January 28 regarding voice cloning. I
think that it is important that all of us are participating,
becoming better educated, and helping make sure we are taking
steps as we move forward.
Clearly, this is a new era, and on one hand we can
celebrate that America has led the world in innovation and
technology and improving our lives in many ways. There is also
this other side that we need to be looking at and making sure
that we are taking the appropriate steps to keep people safe
and secure.
So we will continue this important discussion and continue
to become better educated. Today's hearing was a great part of
that. Thank you, Chair.
Ms. Schakowsky. Thank you very much.
I would like to insert into the record the--I seek
unanimous consent to enter the following documents into the
record: a letter from the SAG-AFTRA, a letter from R Street, a
paper written by Jeffrey Westling of the R Street Institute, a
report from the ATHAR Project on Facebook. And so I seek
unanimous consent.
Without objection, so ordered.
[The information appears at the conclusion of the
hearing.\1\]
---------------------------------------------------------------------------
\1\ The Westling paper and the ATHAR report have been retained in
committee files and also are available at https://docs.house.gov/
Committee/Calendar/ByEvent.aspx?EventID=110351.
---------------------------------------------------------------------------
Ms. Schakowsky. So let me thank all of our witnesses today.
We had good participation from Members despite the fact that
there were other hearings going on.
I remind Members that, pursuant to committee rules, they
have 10 business days to submit additional questions for the
record to be answered by the witnesses, and hopefully in a
reasonably short time. We hope that there will be prompt
answers.
And at this time, the subcommittee is adjourned.
[Whereupon, at 1:00 p.m., the subcommittee was adjourned.]
[Material submitted for inclusion in the record follows:]
[GRAPHICS NOT AVAILABLE IN TIFF FORMAT]
[Dr. Donovan did not answer submitted questions for the
record by the time of publication.]
[GRAPHICS NOT AVAILABLE IN TIFF FORMAT]
[all]