[House Report 116-657]
[From the U.S. Government Publishing Office]
Union Calendar No. 541
116th Congress } { Report
HOUSE OF REPRESENTATIVES
2d Session } { 116-657
_______________________________________________________________________
HIJACKING OUR HEROES: EXPLOITING VETERANS THROUGH DISINFORMATION ON
SOCIAL MEDIA
__________
R E P O R T
of the
COMMITTEE ON VETERANS' AFFAIRS
together with
DISSENTING VIEWS
December 16, 2020.--Committed to the Committee of the Whole House on
the State of the Union and ordered to be printed
______
U.S. GOVERNMENT PUBLISHING OFFICE
19-006 WASHINGTON : 2020
LETTER OF TRANSMITTAL
----------
House of Representatives,
Committee on Veterans' Affairs,
Washington, DC, December 9, 2020.
Hon. Cheryl L. Johnson,
Clerk of the House of Representatives,
Washington, DC.
Dear Ms. Johnson: On December 2, 2020, by voice vote, I was
directed by the Committee on Veterans' Affairs to submit
herewith the Committee's report entitled, ``Hijacking our
Heroes: Exploiting Veterans Through Disinformation on Social
Media.'' Dissenting views are included.
Mark Takano,
Chairman.
Spoofing is a unique and growing threat from foreign actors
targeting our veterans on social media in order to steal their
voices, whether for spreading disinformation and political
propaganda, luring unsuspecting Americans into romance scams,
or simply engaging in commercial fraud, these predators are all
trying to impersonate veterans or veteran service
organizations.
--Chairman Mark Takano
The evidence is clear that veterans have their identity
misappropriated and that they, like other social media users,
could be targets for propaganda or scams.
--Ranking Member Dr. David P. Roe
C O N T E N T S
----------
Page
Executive Summary............................................... 1
Veterans are Specifically Targeted for Spoofing............. 2
Spoofing of Veterans Threaten U.S. Elections................ 3
Action by Law Enforcement and Social Media Platforms Is
Inadequate................................................. 3
The Federal Government and the Social Media Platforms Should
Take Additional Action..................................... 5
Introduction to Spoofing........................................ 5
What is Spoofing?........................................... 5
How is Spoofing Detected?................................... 7
Cheapfakes and Deepfakes.................................... 9
How Spoofing Affects Veterans............................... 12
Political Propaganda & Disinformation................... 12
Commercial Fraud & Scams................................ 19
What Spoofing Looks Like.................................... 21
Political Propaganda & Socially Divisive Content........ 21
Commercial Fraud........................................ 25
A Spoofing Case Study--Vietnam Veterans of America.......... 26
VVA Encounters Challenges to Take Down Spoofed Site..... 27
Growth of Spoofed Site.................................. 27
Conclusion of VVA Investigation......................... 28
Scope of the Spoofing Problem............................... 29
The Social Media Platforms...................................... 30
Facebook.................................................... 30
How Facebook is Combatting Spoofing..................... 30
Is Facebook Doing enough?............................... 37
Twitter..................................................... 39
How Twitter is Combatting Spoofing...................... 39
Is Twitter Doing Enough?................................ 43
The Role of Law Enforcement--Briefing with Committee............ 45
Threat Evaluation and Statistics............................ 46
Communications and Data Sharing............................. 47
Is Law Enforcement Doing Enough?............................ 49
Conclusion...................................................... 50
Recommendations............................................. 53
Union Calendar No. 541
116th Congress } { Report
HOUSE OF REPRESENTATIVES
2d Session } { 116-657
======================================================================
HIJACKING OUR HEROES: EXPLOITING VETERANS THROUGH DISINFORMATION ON
SOCIAL MEDIA
_______
December 16, 2020.--Committed to the Committee of the Whole House on
the State of the Union and ordered to be printed
_______
Mr. Takano, from the Committee on Veterans' Affairs,
submitted the following
R E P O R T
together with
DISSENTING VIEWS
Executive Summary
The threat of foreign individuals and organizations
influencing United States (U.S.) elections by manipulating
social media has been a persistent and growing issue since
before the 2016 election year. The threat was a significant
concern during the 2020 elections.
Recent investigations and analysis document the broad
proliferation of online influence campaigns that originate
overseas. This includes the use of ``spoofing,'' or the act of
disguising an electronic communication from an unknown source
as being from a known, trusted source. A subset of these
operations target the veteran and military service member
communities in order to misappropriate their voices, authority
and credibility. The pervasiveness of social media, as well as
the nature of the specific threat to our election integrity and
the sowing of political discord makes this a critical issue
affecting both veterans and those who value veterans' voices.
As described by Chairman of the House Committee on Veterans'
Affairs, Mark Takano (D-CA), ``the issue of protecting our
elections from foreign influence is one of critical importance
to all Americans and preserving the power of veterans' voices
should be of equal concern to us all.''\1\
---------------------------------------------------------------------------
\1\Hijacking Our Heroes: Exploiting Veterans through Disinformation
on Social Media Before the H. Comm. On Veterans' Affairs, 116th Cong.
at 5 (2019) (hearing transcript) [hereinafter HVAC Committee Hearing
Transcript].
---------------------------------------------------------------------------
VETERANS ARE SPECIFICALLY TARGETED FOR SPOOFING
On Wednesday, November 13, 2019, the House Committee on
Veterans' Affairs held an investigative hearing to examine the
nature and scope of threats posed to the veterans' community
through ``internet spoofing.'' Experts testified that stolen,
misappropriated, or fraudulently created social media accounts
can be used to target veterans for the purposes of
disseminating political propaganda and fake news in order to
influence elections. The witnesses also described romance scams
and commercial fraud being perpetrated using spoofing
techniques. Representatives of three major social media
platforms--Facebook, Instagram, and Twitter--discussed how they
are addressing this threat, particularly considering the 2020
elections, and described best practices for information
sharing, protective measures, and law enforcement cooperation.
The Committee later held a briefing on January 14, 2020, with
representatives from several components of the Federal Bureau
of Investigation (FBI) that handle law enforcement for online
crimes.
Ranking Member Dr. David P. Roe (R-TN) noted during the
hearing, ``The evidence is clear that veterans have their
identity misappropriated and that they, like other social media
users, could be targets for propaganda or scams.''\2\ Although
everyone who uses the internet is subject to online scams,
spamming, phishing, identity theft, and other such risks,
veterans are particularly susceptible to internet spoofing
based on their higher propensity for political engagement
(including running for office, volunteering, and sharing
political opinions and information).\3\ For the purposes of
disseminating political propaganda or exerting influence on
dividing Americans on sensitive political ``wedge issues,''
veterans are targeted because of their close identification
with strong national security policies, patriotism, personal
sacrifice, and honor.\4\ Chairman Takano stated during the
hearing, ``By impersonating veterans, these foreign actors are
effectively eroding the hard-earned power and integrity of
veterans'' voices.''\5\
---------------------------------------------------------------------------
\2\Id. at 9.
\3\John D. Gallacher, Vlad Barash, Philip N. Howard & John Kelly,
Computational Propaganda Research Project, Junk News on Military
Affairs and National Security: Social Media Disinformation Campaigns
Against US Military Personnel and Veterans at 1 (2017), http://
comprop.oii.ox.ac.uk/research/working-papers/vetops/.
\4\Id.
\5\HVAC Committee Hearing Transcript, at 4.
---------------------------------------------------------------------------
Veterans are more likely to be engaged in their
communities, be perceived as leaders, and can exert influence
on political matters (particularly with respect to defense and
national security matters).\6\ Therefore, a successful spoofing
scam that results in a veteran or Veteran Service Organization
(VSO) unknowingly distributing or endorsing a piece of
disinformation can yield greatly increased, and sometimes even
exponential, results due to the added credibility imparted to
that disinformation by virtue of its approval by the veteran or
VSO. With each successive endorsement or share, the credibility
of the disinformation snowballs. The collective association
with actual veterans and VSOs makes it increasingly unlikely
that the disinformation will be closely scrutinized,
questioned, or eventually exposed as fraudulent or misleading.
Moreover, scammers also try to spoof veterans to gain leverage
over them. Many veterans move into jobs requiring security
clearances or within the federal government after they leave
the military--those positions can be jeopardized if the veteran
is compromised through financial fraud, identity theft, or
otherwise becomes susceptible to blackmail.\7\
---------------------------------------------------------------------------
\6\Gallacher et al., Computational Propaganda Research Project,
Junk News on Military Affairs and National Security: Social Media
Disinformation Campaigns Against US Military Personnel and Veterans at
1 (2017), http://comprop.oii.ox.ac.uk/research/working-papers/vetops/.
\7\Kristofer Goldsmith, Vietnam Veterans of America, An
Investigation Into Foreign Entities Who Are Targeting Servicemembers
and Veterans Online at 12-13 (2019), https://vva.org/wp-content/
uploads/2019/09/VVA-Investigation.pdf [hereinafter VVA Report].
---------------------------------------------------------------------------
SPOOFING OF VETERANS THREATEN U.S. ELECTIONS
Internet spoofing became a visible problem in the context
of the 2016 U.S. election, when foreign disinformation spread
widely across social media, including Facebook, Instagram,
Twitter and YouTube, among others. However, disinformation on
social media and information operations conducted by
sophisticated actors have occurred for far longer. In the past
few years, foreign information operations have targeted
divisive political issues within American society and have
sought to manipulate and divide political and social
communities. Unfortunately, our military and veterans'
communities are no exception. Moreover, the incidents of
foreign spoofing increased following the 2016 election, and
industry experts project that these numbers will continue to
increase through 2020 and beyond. Russia's Internet Research
Agency (IRA), a Russian company which has engaged in online
influence operations, more commonly known as a ``troll farm,''
dramatically expanded its information operations after the 2016
U.S. Presidential elections, both in terms of volume and
intensity. Russia and Iran are the most prominent state actors
in this context, but recent work has identified additional
state actors, such as China and Saudi Arabia, using information
operations to target communities and topics of interests.
The Senate Select Committee on Intelligence published a
five-volume bipartisan report focused on Russia's influence
operations. The second volume focused on Russia's use of social
media platforms to influence the election, while the third
volume focused on the shortcomings of Obama Administration
efforts to combat the ongoing attacks. The third volume
highlighted the lack of legislative or regulatory action to
combat a known threat emanating from Russia and its
intelligence services. The Senate Report sheds light on the
broader issues of misinformation campaigns and predatory
schemes targeting veterans presented in a report prepared by
the Vietnam Veterans of America (VVA).
ACTION BY LAW ENFORCEMENT AND SOCIAL MEDIA PLATFORMS
IS INADEQUATE
Industry analysts, journalists, and law enforcement agree
that the problems of internet spoofing and foreign influence
exerted through social media continue to grow at an alarming
pace. However, neither the major platforms nor the FBI were
able to identify an obvious or comprehensive solution to this
ongoing problem. Both continue to devote significant resources
towards combatting spoofing. However, the foreign entities who
perpetrate much of this illicit activity are becoming more
sophisticated in their schemes and are targeting broader swaths
of internet users to more quickly and efficiently disseminate
their fraudulent messaging before they are identified and
deactivated.
Facebook and Twitter note that automated systems can
struggle to differentiate authentic images and accounts from
fraudulent, unauthorized, or duplicated accounts and thereby
risk erroneously flagging and removing legitimate accounts. The
platforms have chosen to err on the side of minimizing false
negatives by relying upon patterns of suspicious activity and
certain tactics or techniques, rather than on other identifying
data (e.g., duplicative names or images, geolocation
information, or ostensible organizational affiliations).
Suspicious activity patterns, such as irregular, repetitive, or
voluminous posting, triggers additional layers of review,
including an examination of the geolocation data in order to
assess where the suspicious activity may be originating. The
final review and removal decisions sometimes warrant human
examination, but often removals are made without any human
review. Although these layered review processes may be
effective in protecting legitimate users, they undoubtedly also
add a significant gap in removal time for fraudulent accounts,
which provides a window within which spoofers can continue to
operate.
Law enforcement agencies, such as the FBI, are constrained
in their abilities to efficiently identify and eliminate
spoofers because the agencies only have limited access to the
data held by the social media platforms. Often these agencies
do not receive important information until after the platforms
have already removed a spoofed account, at which point law
enforcement is unable to actively monitor and trace the account
in real time.
The ability of spoofers to operate from overseas,
anonymously, or by using fraudulent or concealed identities
requires law enforcement to rely upon account identification
data and detailed activity patterns in order to accurately
identify or locate the potential spoofer. However, Title II of
the Electronic Communications Privacy Act (ECPA) (18 U.S.C.
Sec. Sec. 2701-2713), known as the Stored Communications Act,
requires a government entity to serve a subpoena on social
media platforms to compel the production of certain relevant
information. Requiring a time-consuming legal process to obtain
identification data hampers the ability of law enforcement to
respond quickly or to fully understand the scope of a potential
spoofing campaign. Therefore, the law enforcement agencies
recommend increasing the amount and level of detail that the
platforms can easily provide to the authorities.
Past attempts to address this problem have been piecemeal
in nature and have proven ineffective to date. This fragmented
approach has prevented any wholesale, systemic efforts to
tighten rules or law enforcement protocols. Incremental
adjustments have been made by individual platforms, which
leaves an irregular landscape where motivated, corrupt actors
may still be able to exploit weaknesses among the platforms.
THE FEDERAL GOVERNMENT AND THE SOCIAL MEDIA PLATFORMS SHOULD TAKE
ADDITIONAL ACTION
Based on discussions with representatives of law
enforcement, and considering the issues raised by the social
media platforms during the hearing, the Committee believes that
there are additional measures needed to address the growing
threats posed by spoofing. Our recommendations fall into two
broad categories.
The first category is oriented at users of social media and
is defensive in nature, such as teaching users how to be aware
of the dangers posed by spoofers on social media and training
them how to protect themselves through heightened vigilance,
healthy skepticism, and adherence to basic principles of cyber-
hygiene.
1. Improve Awareness through a Public Service
Announcement Campaign
2. Develop Cyber-hygiene Training
3. Strengthen Partnership Between Social Media
Platforms and VSOs
The second category is aimed at putting the social media
platforms and law enforcement on the offensive and developing
robust mechanisms to more effectively identify and quickly
eliminate foreign-based spoofers. While the first category is
likely to be less costly and easier to implement, the second
category may ultimately prove to be more effective in bringing
the threat under control.
4. Improve Reviews of Accounts by Social Media
Platforms
5. Consider Legislative Reforms to Facilitate Sharing
Information
6. Increase Data Sharing on Fraudulent Accounts
7. Improve Identity Verification and Geolocation
Identification
Introduction to Spoofing
Veterans and the veterans' community have consistently been
targeted by scammers and malicious actors seeking to exploit
their valor, prestige, and assets. Since the advent of the
internet, new types of risks for scams, misinformation, and
fraud have become prevalent. Spoofing now represents the latest
tactic used by bad actors to try and target veterans and their
supporters.
WHAT IS SPOOFING?
In its simplest terms, ``spoofing'' is the act of
disguising an electronic communication from an unknown source
as being from a known, trusted source--either by creating a
fraudulent account, or by hijacking a real account.\8\
Websites, Facebook pages, Twitter accounts, and other social
media can all be spoofed by bad actors seeking to deceive or
trick unsuspecting viewers and are referred to as spoofed
websites, spoofed pages, or spoofed accounts. While all users
of the internet are generally subject to the potential risks of
fraud, deception, and theft, spoofing on social media is often
specifically targeted at particular groups or types of users.
This includes, notably, veterans. Veterans and VSOs are being
targeted for internet spoofing scams, in which social media
accounts and online profiles are being stolen, hijacked,
fraudulently created, copied, or otherwise faked to
misappropriate veterans' identities, voices, images, and
affiliations.\9\
---------------------------------------------------------------------------
\8\Alex Horton, Russian Trolls Are Targeting American Veterans, and
Trump's Government Isn't Helping, Group Says, The Washington Post (Jan.
7, 2020), https://www.washingtonpost.com/national-security/2020/01/07/
russian-trolls-are-targeting-american-veterans-trumps-government-isnt-
helping-group-says/.
\9\There are four different types of Veteran Service Organizations:
(1) Congressionally chartered Veterans Service Organizations that are
also recognized by the Department of VA Office of General Counsel for
the purpose of preparation, presentation, and prosecution of claims
under laws administered by the Department of Veterans Affairs, as
provided in Section 5902 (formerly Section 3402) of Title 38, United
States Code (U.S.C.) and Sub Section 14.628(a) and (c) of 38 C.F.R.,
(2) Congressionally chartered Veterans Service Organizations but that
are NOT recognized by the Department of Veterans Affairs for the
purpose of preparation, presentation and prosecution of Veteran's
claims only, (3) Veteran organizations NOT congressionally chartered
but are officially recognized by the Department of Veterans Affairs for
the purpose of preparation, presentation and prosecution of Veteran's
claims only, and (4) Veteran organizations not congressionally
chartered or officially recognized by the Department of Veterans
Affairs for the purpose of preparation, presentation and prosecution of
Veteran's claims only. Additionally, there are VSOs that are
categorized as Intergovernmental Affairs organizations. See generally
U.S. Dep't of Veterans Affairs, Veterans and Military Service
Organizations and State Directors of Veterans Affairs (2019), available
at https://www.va.gov/vso/VSO-Directory.pdf.
---------------------------------------------------------------------------
Internet spoofing can take many different forms and be
deployed for a wide range of nefarious behavior with
significant and damaging results to individual veterans and
their families, and even to our national security and election
integrity. Much, but not all, of this fraudulent online
activity is perpetrated by foreign actors, and even in some
cases, by state-backed foreign actors.\10\ Through such online
misappropriation, veterans' images, identities, and voices are
being illicitly used to influence our elections by
disseminating political propaganda, disinformation and fake
news. Spoofers also misappropriate veterans' images and stories
in order to conduct romance scams and engage in various other
forms of commercial fraud.\11\
---------------------------------------------------------------------------
\10\S. Rep. No. 116-XX, Volume 2 at 11 (2019).
\11\ Darla Mercado, These Scammers Have Set Their Sights on Members
of the Military, CNBC (Nov. 13, 2019), https://www.cnbc.com/2019/11/13/
these-scammers-have-ripped-off-405-million-from-members-of-the-
military.html.
---------------------------------------------------------------------------
Spoofed accounts can often be identified by certain
patterns of posting activity, growth rates of the follower and
subscriber base, or signs of foreign control. Kristofer
Goldsmith, the Founder and President of High Grounds Veterans
Advocacy, and the former Associate Director of Policy &
Government Affairs for the Vietnam Veterans of America (VVA), a
congressionally chartered VSO, conducted an investigation into
spoofing after discovering that VVA itself had been spoofed.
Through the course of investigating the VVA spoof, VVA learned
that many spoofed websites commonly feature high numbers of
followers or subscribers, irregular, repetitive, or voluminous
posting activity, and often have foreign-based administrators.
Goldsmith provides the following example in the VVA report:
One such page, ``Veterans of Vietnam,'' with nearly
160,000 followers, has had admins in Russia, Ukraine,
and Italy. This page has been bolstered by at least
three dedicated Russian-generated Vietnam-veteran
focused websites that were created to build the
Facebook page's credibility by sharing information
about the Vietnam War and veterans' benefits. These
admins also control a closed Facebook group, ``American
Veterans of Vietnam,'' which solicits information from
Vietnam veterans regarding their military experience.
Fake accounts are also being utilized by hostile
Chinese intelligence services to connect with high-
ranking and influential members of the intelligence and
defense communities centered in and around Washington,
DC. Chinese officials are seeking to exploit
financially vulnerable members of these communities and
leverage debts to recruit spies.\12\
---------------------------------------------------------------------------
\12\Goldsmith, VVA Report, at 137.
Spoofing on social media platforms can be as simple as
creating a social media account using a name, image, or
affiliation that is not owned by or authorized for the creator.
Spoofing does not require ``hacking'' an existing account or
gaining access to a password. Instead, spoofing generally
involves the creation of a new account that fraudulently
purports to be the account of an individual or entity with
which it has no actual connection. These fake sites then
rapidly build up a dedicated following by disseminating
carefully curated memes (captioned pictures, GIFs, or videos,
often altered to be humorous, that are copied and spread online
in a viral manner),\13\ images, propaganda, and fake or
modified news stories, all of which are deliberately designed
to provoke an emotional response from a targeted group,
accompanied by sharing, liking, endorsing, and following the
fake group and its content.\14\ This content often involves
copying publicly available images, recycling old news stories
with minor modifications or repurposing outdated stories to
leverage the changed circumstances to dupe unsuspecting
readers. Spoofers deliberately leverage emotionally sensitive
topics, often involving politics or divisive social issues, by
using simplistic memes or manipulative images to elicit a
strong reaction and reflexive ``like'' or ``share.''\15\
---------------------------------------------------------------------------
\13\Merriam-Webster, Meme Definition, https://www.merriam-
webster.com/dictionary/meme.
\14\Goldsmith, VVA Report, at 25.
\15\ Id.
---------------------------------------------------------------------------
HOW IS SPOOFING DETECTED?
Internet spoofing must be detected by examining repeated
patterns of suspicious account activity or online behavior, as
opposed to a simple inspection of the names, images, or
purported identities associated with a given account. Social
media platforms are generally ill-equipped to determine whether
a given image or association is authentic or authorized, so
when a duplicative account is brought to their attention, it is
not readily apparent which account is the real one and which is
the fake. The use of stock photographs, celebrity images,
sports team logos, and alternative spelling/nicknames further
complicates the ability of a social media platform to identify
fraudulent or misappropriated accounts efficiently or
accurately based solely on the identifying criteria associated
with the account. Moreover, many users may have multiple
accounts on each social media platform for legitimate purposes,
such as separating personal and professional accounts, or
maintaining independent family and personal accounts. A simple
review of basic account information is insufficient to enable
the social media platform to reliably make any determinations
about which of these accounts were legitimate. Therefore, the
platforms examine account behavior and patterns of activity to
identify potentially suspicious trends which may indicate a
spoofed or fraudulent account.
Using automatic machine review and artificial intelligence
to rapidly analyze large volumes of internet traffic enables
the platform to identify patterns of account activity that do
not fit within projected norms of standard behavior. Examples
of these patterns include posting at unusual rates, times, or
volumes, repeated posting or sharing of the same content, or
near instantaneous posting or commenting on particular pages
that may indicate automated posting, often referred to as ``bot
activity.''\16\ Some specific types of signals that may
indicate ``suspicious activity'' include coordinated
inauthentic behavior such as posting identical content on
different platforms or pages nearly simultaneously; spelling
and grammatical mistakes potentially indicative of non-native
English speakers; distributing URLs that are associated with
malware; masking the actual identity of links by using URL-
shorteners; soliciting personal information; the use of
advertising tools to target and retarget specific users (such
as veterans); and the use of duplicative images, memes, or
links across multiple accounts and platforms. Suspicious
activity can also include the dissemination of foreign-state-
sponsored and state-controlled propaganda from known sources
such as TASS, RT, and Sputnik News.\17\ Other indicia of
suspicious activity may include altering names and themes of
pages and groups related to veterans, and the false
representation of veteran status or VSO affiliation.\18\
---------------------------------------------------------------------------
\16\ A bot is a computer program that performs automatic repetitive
tasks. Merriam-Webster, Bot Definition, https://www.merriam-
webster.com/dictionary/bot.
\17\Goldsmith, VVA Report, at 14.
\18\Id.
---------------------------------------------------------------------------
The platforms regularly use a combination of human
reviewers and artificial intelligence to screen and review
content for certain violations, such as pornography, violent
images, and some intellectual property violations. The specific
nature of the questionable content in spoofing makes it
particularly challenging for artificial intelligence alone to
identify and verify material from fraudulent accounts.
Similarly, human fact-checkers can screen factual
disinformation, but can struggle to independently and
efficiently verify photographs or associated identities, among
others. By supplementing human review with artificial
intelligence, the platforms have had some success in detecting
behaviors that are difficult for bad actors to fake, including
connections to others on the platform. For example, in December
2019, Facebook removed hundreds of accounts across all of its
platforms, including Instagram, that were associated with a
group that had used AI-generated profile pictures to pose as
real people and then spread misinformation through the
resulting artificially-expanded networks.\19\ This marked the
first reported instance that AI-generated user profiles
launched at scale and used in an influence operation social
media campaign were identified and removed.\20\ Twitter also
removed hundreds of fraudulent accounts generated by the same
group as part of a coordinated global spoofing campaign.\21\
---------------------------------------------------------------------------
\19\Tony Romm & Isaac Stanley-Becker, Facebook, Twitter Disable
Sprawling Inauthentic Operation that Used AI to Make Fake Faces, The
Washington Post (Dec. 20, 2019), https://www.washingtonpost.com/
technology/2019/12/20/facebook-twitter-disable-sprawling-inauthentic-
operation-that-used-ai-make-fake-faces/.
\20\Id.
\21\Id.
---------------------------------------------------------------------------
CHEAPFAKES AND DEEPFAKES
The challenge of disinformation is magnified when accounts
post materials that do not objectively violate the terms of
service, but instead make false claims or distribute doctored
media, including ``cheapfake'' or ``deepfake'' videos, or
create accounts using pictures generated by artificial
intelligence in order to quickly and surreptitiously build a
massive global network.
Cheapfakes use conventional techniques like speeding,
slowing, cutting, re-staging, or re-contextualizing footage to
alter how the media is widely perceived.\22\ The use of
photoshopping and lookalikes are common cheapfake methods used
to create doctored images to circulate through the media. An
easy mode of producing a cheapfake is simply cutting together
existing footage, speeding up or slowing down that footage, or
altering the audio and spreading it under false pretenses. This
threat looms large because cheapfakes are easy to make and
distribute through powerful social media platforms designed to
spread engaging content widely and quickly.
---------------------------------------------------------------------------
\22\Britt Paris & Joan Donovan, Data & Society, Deepfakes and Cheap
Fakes: The Manipulation of Audio and Visual Evidence 23-32 (2019),
https://datasociety.net/wp-content/uploads/2019/09/
DS_DeepfakesDS_CheapDS_FakesFinal-1-1.pdf [hereinafter Deepfakes &
Cheap Fakes].
---------------------------------------------------------------------------
Deepfake media content, on the other hand, is audio or
video that has been fabricated with very sophisticated tools to
make someone appear to say or do something they did not really
do--from harmless satire to propaganda--and are increasingly
difficult to differentiate from legitimate media.\23\ Doctored
cheapfake clips have been used for decades to distort viewers'
reactions, including a slowed video showing LAPD officers
beating Rodney King that was used by the officers' defense
counsel during their 1993 trial in order to sow doubt with the
jury, to a more recent May 2019 example, when a manipulated
video clip was decreased in speed by almost 75% in order to
depict Nancy Pelosi ``drunkenly'' slurring her words while
talking about Donald Trump.\24\ The video of Speaker Pelosi was
altered to give the impression that her speech was sluggish,
suggesting perhaps that she had a mental ailment or that she
was drunk.\25\ This video had more than 11 million views in
just five days. Notably, YouTube removed this video from its
platform as a violation of its policy on technical manipulation
of videos.\26\ Twitter did not remove the video immediately,
and did not comment to explain its decision.\27\ Facebook
declined to remove the video, even after its third-party fact-
checking partners deemed the video to be false, stating
instead, ``We don't have a policy that stipulates that the
information you post on Facebook must be true.''\28\
---------------------------------------------------------------------------
\23\Drew Harwell, Top AI Researchers Race to Detect `Deepfake'
Videos: `We Are Outgunned', The Washington Post (June 12, 2019),
https://www.washingtonpost.com/technology/2019/06/12/top-ai-
researchers-race-detect-deepfake-videos-we-are-outgunned/.
\24\Paris & Donovan, Deepfakes & Cheap Fakes, at 30.
\25\ Drew Harwell, Faked Pelosi Videos, Slowed to Make Her Appear
Drunk, Spread Across Social Media, The Washington Post (May 24, 2019),
https://www.washingtonpost.com/technology/2019/05/23/faked-pelosi-
videos-slowed-make-her-appear-drunk-spread-across-social-media/.
\26\Greg Bensinger, As Primary Season Gets Underway, YouTube Cracks
Down on Doctored Election Videos, The Washington Post (Feb. 3, 2020),
https://www.washingtonpost.com/
technology/2020/02/03/youtube-election-videos/.
\27\Drew Harwell, Facebook Acknowledges Pelosi Video Is Faked but
Declines to Delete It, The Washington Post (May 24, 2019), https://
www.washingtonpost.com/technology/2019/05/24/facebook-acknowledges-
pelosi-video-is-faked-declines-delete-it/.
\28\Id.
---------------------------------------------------------------------------
Each of these technical tool sets was previously only
available to experts, but with technological advancement and
widespread social media use, these are more accessible to
amateurs and their outputs reach larger scales at higher
speeds. Today, social media is experiencing a rapid increase of
image and video distribution and redistribution. Members of the
public have the ability to spread messages at a larger scale,
with less oversight than ever before.
For example, deepfake technology has already been deployed
in India's 2020 elections. There, a candidate recorded a video
in English and then used deepfake technology to change his
speech and the specific movement of his mouth to make it appear
that he was speaking in Hindi in a strategic effort to solicit
Hindi-speaking voters.\29\ While this use was not necessarily
for nefarious purposes, the growing ease of creating such
convincing videos demonstrates the imminent risks and dangers
that are on the doorstep. Acknowledging the imminent threat
posed by deepfakes and manipulated videos, YouTube--the largest
purveyor of video content and the world's second largest search
engine--has announced a policy banning technically manipulated
or doctored content that may mislead viewers.\30\ This policy
excludes video clips simply taken out of context without
further manipulation. Twitter also has a policy prohibiting
synthetic and manipulated media, including deepfakes and
cheapfakes.\31\ Facebook has implemented a similar policy, but
with significant exclusions as discussed below.\32\
---------------------------------------------------------------------------
\29\Charlotte Jee, An Indian Politician Is Using Deepfake
Technology to Win New Voters, MIT Technology Review (Feb. 19, 2020),
https://www.technologyreview.com/f/615247/an-indian-politician-is-
using-deepfakes-to-try-and-win-voters/.
\30\ Leslie Miller, How YouTube Supports Elections, Youtube:
Official Blog (Feb. 3, 2020) (Miller is the VP of Government Affairs
and Public Policy), https://youtube.googleblog.com/2020/02/how-youtube-
supports-elections.html.
\31\Twitter, Synthetic and Manipulated Media Policy, https://
help.twitter.com/en/rules-and-
policies/manipulated-media.
\32\Facebook, Manipulated Media Policy, https://www.facebook.com/
communitystandards/
manipulated_media.
---------------------------------------------------------------------------
Deepfake technology poses two parallel sets of problems.
First, deepfake technology could in the future be capable of
producing fraudulent content of such high quality that current
detection methods will be unable to evaluate the legitimacy of
the material.\33\ Second, as deepfake media becomes more
interspersed with authentic media, more people will tend to
ignore or dismiss legitimate news. This is particularly
dangerous for the veteran population whose aging demographic
may be less familiar with newer technology than younger
groups.\34\
---------------------------------------------------------------------------
\33\A study conducted by Deeptrace, an Amsterdam-based
cybersecurity company, revealed that in 2019 there were approximately
15,000 deepfake videos, nearly double the amount online 2018. Of these
15,000 videos 96 percent were pornographic in nature and the top four
most visited deepfake pornography websites had over 134 million views.
Additionally, the report found that a marketplace for deepfake creators
to sell their services has cropped up--selling services, including
``bespoke faceswap videos for $30 to custom voice cloning for $10 per
50 words generated.'' Henry Ajder, Giorgio Patrini, Francesco Cavalli &
Laurence Cullen, Deeptrace, The State of Deepfakes: Landscape, Threats,
and Impact (2019), https://regmedia.co.uk/2019/10/08/
deepfake_report.pdf.
\34\David Frank, Veterans Twice as Likely to Be Scammed, AARP:
Scams & Fraud (Nov. 8, 2017), https://www.aarp.org/money/scams-fraud/
info-2017/veterans-scam-protection-fd.html.
---------------------------------------------------------------------------
Compounding the emerging threat posed by cheapfakes and
deepfakes are the particular definitions of ``manipulated
video'' used by platforms to establish and regulate their own
content standards and community guidelines. For example,
Facebook's policy banning manipulated video requires both 1)
that the video be edited, synthesized, and likely to mislead,
and 2) that the video be the product of artificial
intelligence, machine learning, or deep learning, that merges,
combines, replaces, and/or superimposes content onto a video,
creating a video that appears authentic.\35\ Moreover, and
perhaps most troubling, Facebook's policy also excludes content
that has been edited to omit words that were said or change the
order of words that were said.\36\ Selectively removing or
resequencing words can obviously lead to fundamentally
different meanings than what the speaker intended, and so it is
unclear how or why this exclusion is consistent with Facebook's
stated objective of reducing misleading content. Even beyond
spoken words, Facebook has also allowed content that splices
and reorganizes actions in a video to deliberately portray a
different sequence of events than what actually occurred. For
example, a cheapfake video involving Speaker Nancy Pelosi was
shared widely following the State of the Union address on
February 4, 2020. This cheapfake was strategically edited to
juxtapose her ripping up the President's speech with his
comments recognizing veterans and a redeployed soldier
reuniting with the soldier's family, when in reality she had
torn up the speech after the President had concluded his
remarks.\37\ This video was shared to allude that the Speaker
was disrespecting veterans and the central issues faced by
veterans. This type of misinformation parallels other schemes
which use veterans and veteran issues as wedges to deepen the
divide between political parties and paint one party or the
other as ``anti-veteran.''
---------------------------------------------------------------------------
\35\Facebook, Manipulated Media Policy, https://www.facebook.com/
communitystandards/
manipulated_media.
\36\Id.
\37\Queenie Wong, Facebook, Twitter Called on to Ax Edited Clip of
Pelosi Tearing Trump Speech, CNET (Feb. 7, 2020), https://www.cnet.com/
news/facebook-twitter-under-pressure-to-
remove-edited-video-of-pelosi-ripping-up-trumps-speech/.
---------------------------------------------------------------------------
Spoofing should be differentiated from stealing or
hijacking control of a legitimate account. Spoofing is
predicated upon fraud, such that the victim mistakenly believes
the spoofed account to belong to a person or organization, but
which actually has no connection to that known entity. Stolen
or hijacked accounts, on the other hand, are the authentic
accounts of the named person or organization, which are being
controlled by an unauthorized person, without the knowledge or
permission of the legitimate owner. Therefore, many of the
standard cyber-security protocols intended to protect internet
users from phishing,\38\ data breaches, and compromised
accounts are not as effective in the spoofing context, although
educating users to the importance of safeguarding personal data
and being cautious when entering into any financial
transactions is always valuable. While spoofing or
impersonation is broadly against the Terms of Service for the
major social network platforms, it is unclear whether simple
spoofing, short of any commercial or financial fraud, is
illegal. Once a spoofed account is used to perpetrate financial
fraud (including romance scams) it most likely falls under
federal criminal wire fraud statutes.\39\
---------------------------------------------------------------------------
\38\Phishing is a scam by which an Internet user is duped (as by a
deceptive e-mail message) into revealing personal or confidential
information which the scammer can use illicitly. Merriam-Webster,
Phishing Definition, https://www.merriam-webster.com/dictionary/
phishing.
\39\18 U.S.C.Sec. 1343.
---------------------------------------------------------------------------
HOW SPOOFING AFFECTS VETERANS
While anyone using the internet is subject to the risks of
spoofing, the veterans' community is particularly targeted and
exploited by these scammers as previously discussed. Spoofing
affects veterans in two distinct ways--veterans can be harmed
by spoofing either when they are specifically targeted as the
direct victims of spoofing, or when they are exploited by
spoofers to specifically target a different, often non-veteran,
victim. The former category includes the dissemination of
political propaganda and fake news through spoofed accounts
pretending to be veteran or VSO accounts with the specific
intent of leveraging the influence and authority gleaned from
that false affiliation. The latter category includes the
perpetration of romance scams and other forms of commercial
fraud where the targeted victim is not necessarily a veteran,
but where the targeted victims incorrectly believe themselves
to be interacting with an actual veteran or VSO. In both cases,
the intended victims are targeted through the misuse of images,
memes, fake news stories, and other disinformation transmitted
via misappropriated, stolen, or fraudulent social media
accounts.
A common element of these types of spoofing schemes is the
misappropriation of veterans' voices to gain public trust. As
Chairman Takano stated, ``Pretending to be a veteran for any
reason is shameful, but it is especially shameful when such
deception is used to spread disinformation.''\40\
---------------------------------------------------------------------------
\40\HVAC Committee Hearing Transcript, at 4.
---------------------------------------------------------------------------
Veterans are also targeted because they can be particularly
susceptible to blackmail or financial leverage if their
personal information is compromised through a spoofing
campaign. Many veterans continue to work in positions involving
national security or otherwise requiring a security clearance,
and any ability by a spoofer to compromise that security
clearance would directly jeopardize the veteran's
employment.\41\ For example, if a veteran becomes ensnared in a
romance scam that results in the veteran's identity being
stolen or their credit ruined, then their security clearance
may be revoked, and they could lose their job. The VVA report
noted, ``Additionally, nearly one-third of the federal
workforce is composed of veterans. This makes the targeting of
the military and veteran population a means to jeopardize
federal agencies ranging from law enforcement and defense to
healthcare and food safety.''\42\
---------------------------------------------------------------------------
\41\Goldsmith, VVA Report, at 12.
\42\Id.
---------------------------------------------------------------------------
Political Propaganda & Disinformation
Veterans carry significant credibility and influence in
their communities, especially on issues related to patriotism,
national security, defense, and public service. Chairman Takano
stated during the hearing, ``Veterans wield considerable
influence and credibility in their communities earned by virtue
of their selfless sacrifice and service to our country.''\43\
Spoofers attempting to spread disinformation or fake news can
leverage that credibility to amplify their messages by posing
as veterans. The VVA report states, ``Foreign adversaries have
many motivations for targeting members of the military and
veteran community. This population has a higher propensity than
other subgroups of Americans to be politically engaged--they
are more likely to vote and serve in public office--and they
tend to wield greater political influence on those around
them.''\44\ Assuming the identity of a VSO or an individual
veteran imparts a degree of reliability or authority to
articles or news stories relating to those issues, which in
turn makes that story more likely to be further shared.
Increasing the number of ``likes'' on social media and
spreading the story broadly through repeated sharing are the
twin pillars of disseminating fake news and political
propaganda.
---------------------------------------------------------------------------
\43\HVAC Committee Hearing Transcript, at 4.
\44\John D. Gallacher et al., Computational Propaganda Research
Project, Junk News on Military Affairs and National Security: Social
Media Disinformation Campaigns Against US Military Personnel and
Veterans (2017), http://comprop.oii.ox.ac.uk/research/working-papers/
vetops/.
---------------------------------------------------------------------------
The content to which spoofers generally attempt to affix
the misappropriated veteran endorsement includes socio-
politically divisive issues predicated upon categorizing
veterans, military, law enforcement, and ``patriots,'' defined
broadly, on one side, and thereby positioning the ``others'' on
the opposite side as necessarily unpatriotic, un-American, or
at best, soft on crime or national defense. For example, issues
like immigration policy, Black Lives Matter, or kneeling during
the national anthem have all been used to target veterans and
their associates such as families, friends, supporters, and
affinity groups. Research conducted at the University of
Washington concluded ``that the examined trolling accounts
systematically took advantage of these divisions,'' and
specifically focused on the hashtag BlackLivesMatter.\45\
---------------------------------------------------------------------------
\45\Leo G. Stewart, Ahmer Arif & Kate Starbird, University of
Washington, Examining Trolls and Polarization with a Retweet Network
(2018), https://faculty.washington.edu/kstarbi/examining-trolls-
polarization.pdf.
---------------------------------------------------------------------------
Using veteran affiliated pages, or pages that appear to
have such affiliations, to spread memes and images that
positioned then-candidate Donald Trump as ``pro-military'' or
supportive of veterans and President Barack Obama, Vice-
President Joe Biden, or then-candidate Hillary Clinton as
opposed to veterans and the military served to build and
reinforce a partisan divide. This divide was then further
exploited by the fraudulent veteran pages to spread
disinformation or fake news relating to other issues ranging
from race-baiting to anti-Semitism to Antifa. These fraudulent
pages also often distribute fake news stories, including
stories which resuscitate authentic issues from prior years,
but change the dates to make it falsely appear that a given
candidate or political party is promoting policies that hurt
veterans or soldiers. Disinformation and fake news are also
used by spoofers to target the broader electorate beyond
veterans in order to achieve similar partisan divisions.
The use of disinformation for partisan purposes in a
spoofing operation occurred during the 2017 Alabama Special
Election for U.S. Senate between Republican candidate Roy Moore
and Democratic candidate Doug Jones.\46\ In that race, a
Facebook page named ``Dry Alabama'' became ``the stealth
creation of progressive Democrats who were out to defeat Mr.
Moore.''\47\ The ``Dry Alabama'' page was created by a
Democratic operative named Matt Osborne who intended to spread
false information tying Mr. Moore to a movement for the
prohibition of alcohol in Alabama. The plan was to associate
Mr. Moore with the prohibition effort to hurt his chances of
earning votes from moderate, business-oriented Republican
voters. Mr. Moore was never an advocate of the prohibition
movement, and Mr. Osborne admitted that he had fabricated the
claim.\48\
---------------------------------------------------------------------------
\46\Scott Shane & Alan Blinder, Democrats Faked Online Push to
Outlaw Alcohol in Alabama Race, The New York Times (Jan.7, 2019),
https://www.nytimes.com/2019/01/07/us/politics/
alabama-senate-facebook-roy-moore.html.
\47\Id.
\48\Id.
---------------------------------------------------------------------------
There was also a second known spoofing operation in this
Alabama Senate race, in which a series of fraudulent Twitter
accounts purporting to be Russians were established to follow
Mr. Moore's tweets. This gave the impression that Mr. Moore was
being supported by Russian operatives.\49\ This spoofing
campaign was funded by Democratic billionaire Reid Hoffman,
with the intention of tying Mr. Moore to Russian operatives to
support a parallel effort in which the spoofers would pose
online as conservative Alabamians advocating a write-in
campaign in lieu of voting for Mr. Moore.\50\ The problem of
spoofing is not limited to one side of the political aisle and
its victims are the American people. If individuals or groups
are able to influence elections with false narratives, then
faith in the electoral process is undermined.\51\
---------------------------------------------------------------------------
\49\Id.
\50\Scott Shane & Alan Blinder, Secret Experiment in Alabama Senate
Race Imitated Russian Tactics, The New York Times (Dec. 19, 2018),
https://www.nytimes.com/2018/12/19/us/alabama-senate-roy-jones-
russia.html.
\51\Election infrastructure targeted to ``undermine confidence in
election''. S. Rep. No. 116-XX, Volume 1 at 10 (2019).
---------------------------------------------------------------------------
While the net effect of such spoofing campaigns often has
specific political objectives, the methods and content used to
achieve the requisite divisions in society are often deployed
through facially apolitical or neutral pages (e.g. the fake
Vietnam Vets of America, Veterans Nation, We Are Veterans).\52\
On the other hand, there are also specifically-partisan pages,
like Vets For Trump, that peddle similar content with the same
underlying objective of sowing social divisions through
illegitimate means.\53\ These overtly partisan pages or
websites generally lack any official relationship with the
candidate or party they ostensibly support, and often originate
overseas.\54\ A foreign-based, partisan-identified page
disseminating propaganda or divisive content would be the
clearest example of the threat posed to American election
integrity from foreign spoofers.
---------------------------------------------------------------------------
\52\Goldsmith, VVA Report, at 58.
\53\Id. at 7.
\54\Id.
---------------------------------------------------------------------------
For example, in the spring of 2019, a legitimate American
Facebook page called ``Vets For Trump'' was hijacked by
Macedonian actors for several months.\55\ The legitimate owners
contracted with a Macedonian firm to manage and expand the
page's advertising revenues, and the Macedonian actors
exploited their access to take over complete control of the
page and lock out the American owners.\56\ During this period
of exclusive Macedonian control, the spoofers used PayPal for
fraudulent fundraising, but the PayPal account they used was
tied to a known Macedonian spoofer and had no connections to
legitimate, registered American fundraising entities.\57\ Thus,
unwitting donors who were lured into this site and who
genuinely believed that they were making political
contributions to support President Donald Trump through ``Vets
For Trump'' were actually funding this group of Macedonian
spoofers.\58\ This marks one of the first known instances of
foreign interference in American political activity ahead of
the 2020 election.\59\
---------------------------------------------------------------------------
\55\Id. at 142.
\56\Id. at 145; Craig Timberg, The Facebook Page `Vets for Trump'
Was Hijacked by a North Macedonian Businessman. It Took Months for the
Owners to Get It Back., The Washington Post (Sept. 17, 2019), https://
www.washingtonpost.com/technology/2019/09/17/popular-facebook-page-
vets-trump-seemed-be-place-former-military-months-macedonians-
controlled-it/.
\57\Goldstein, VVA Report, at 149-155.
\58\Id.
\59\Id.
---------------------------------------------------------------------------
The general analysis of foreign influence in the 2016
election identified vulnerabilities and opportunities for
leverage that have not yet been adequately addressed at a
systemic level.\60\ The current approach of ad-hoc review and
removal of violative content by the social media platforms
themselves enables the perpetrators to continue operations by
simply creating new accounts or switching platforms.\61\ The
intelligence community's assessment of the 2016 election was
that foreign actors, primarily Russia's Internet Research
Agency (IRA), were successful in conducting broad
disinformation campaigns across multiple social media platforms
that targeted specific rift lines in the American
electorate.\62\ The Senate Select Committee on Intelligence
published a five-volume bipartisan report focused on Russia's
influence operations.\63\ The second volume focused on Russia's
use of social media platforms to influence the election, while
the third volume focused on the short comings of the Obama
Administration efforts to combat the ongoing attacks. The
overarching theme of this third volume highlighted the lack of
U.S. legislative or regulatory action to combat a known threat
emanating from Russia and its intelligence services.\64\ The
Senate reports shed light on the broader issues presented in
the Vietnam Veterans of America report of misinformation
campaigns and predatory schemes on veterans.
---------------------------------------------------------------------------
\60\Karoun Demirjian & Devlin Barrett, Obama Team's Response to
Russian Election Interference Fell Short, Senate Report Says, The
Washington Post (Feb. 6, 2020), https://www.washingtonpost.com/
national-security/obama-teams-response-to-russian-election-
interference-fell-short-senate-report-says/2020/02/06/93c2fdac-48f2-
11ea-9164-d3154ad8a5cd_story.html.
\61\HVAC Round Table Discussion with FBI on January 14, 2020.
\62\S. Rep. No. 116-XX, Volume 3 (2020).
\63\Id.
\64\Id.; see also Karoun Demirjian & Devlin Barrett, Obama Team's
Response to Russian Election Interference Fell Short, Senate Report
Says, The Washington Post (Feb. 6, 2020), https://
www.washingtonpost.com/national-security/obama-teams-response-to-
russian-election-
interference-fell-short-senate-report-says/2020/02/06/93c2fdac-48f2-
11ea-9164-d3154ad8a5cd_story.html.
---------------------------------------------------------------------------
Russian state-backed GRU disinformation campaigns actually
increased in the aftermath of the 2016 election, according to
the Senate Intelligence Committee report.\65\ Foreign actors
continue to pursue disruption of the American political process
by spreading disinformation, divisive content, or propaganda,
and it is not clear that social media platforms have
sufficiently addressed this threat or prepared their users to
protect themselves. Spoofing and disinformation have continued
to be a present and growing threat over the past three years
and are only likely to increase in an election year.\66\
---------------------------------------------------------------------------
\65\S. Rep. No. 116-XX, Volume 2 at 42 (2019).
\66\Suzanne Spaulding, Jason Gresh, Devi Nair & Alexandra Huber,
Why the Kremlin Targets Veterans, CSIS (Nov. 8, 2019), https://
www.csis.org/analysis/why-kremlin-targets-veterans.
---------------------------------------------------------------------------
The spoofing threat has evolved and expanded since 2016,
with a greater role now played by Instagram and YouTube in the
dissemination of disinformation, memes, and political
propaganda.\67\ As younger users migrate away from Facebook,
these and other emerging platforms are becoming more popular
and influential with that demographic. The visual nature of the
content on both Instagram and YouTube also supports the
dissemination of memes and videos, which are very effective
conduits for disinformation. For example, in the midst of the
coronavirus pandemic, a YouTube video posted by a QAnon
supporter combined numerous false and misleading claims to
suggest that the pandemic was actually a political hoax.\68\
The video subsequently garnered millions of views across
multiple other social media platforms.\69\
---------------------------------------------------------------------------
\67\Paul M. Barrett, NYU Stern Center for Business and Human
Rights, Disinformation and the 2020 Election: How the Social Media
Industry Should Prepare (2019), https://issuu.com/
nyusterncenterforbusinessandhumanri/docs/nyu_election_2020_report?
fr=sY2QzYzI0MjMwMA.
\68\The Associated Press, Video Stitches False Claims Together to
Paint COVID-19 as a Political Hoax, AP News (July 9, 2020), https://
apnews.com/afs:Content:9065413346.
\69\Id.
---------------------------------------------------------------------------
Researchers have also found that the YouTube recommendation
algorithm steers viewers toward increasingly radical and
extreme videos.\70\ Foreign entities may be able to quietly
disseminate disinformation by generating innocuous and popular
content, and then relying on the algorithm to divert viewers or
subscribers to other less innocuous content. For example, the
third largest reach of any entertainment channels on YouTube in
November 2019 (surpassed only by Disney and Warner Media) was
held by TheSoul Publishing--a Cypriot entity run by Russian
nationals, with YouTube and Google advertising revenues of tens
of millions of dollars.\71\ While the vast majority of YouTube
content created and posted by TheSoul Publishing consists of
short, non-political videos related to crafting, hobbies, and
listicles, they also post some videos featuring political and
historical disinformation with strong pro-Russian and anti-
American perspectives.\72\ By accumulating a massive subscriber
base for their various channels, TheSoul Publishing and similar
entities are able to establish a built-in audience to which it
can distribute its disinformation and political content.\73\
TheSoul Publishing has also purchased Facebook advertisements
on political issues targeting American voters, and used rubles
to pay for the ad buys.\74\ The cross-pollination of content
across the various social media platforms further enables the
rapid dispersal of specific messages, particularly memes and
videos.
---------------------------------------------------------------------------
\70\Karen Kornbluh, The Internet's Lost Promise: And How America
Can Restore It, Foreign Affairs (September/October 2018), https://
www.foreignaffairs.com/articles/world/2018-08-14/internets-lost-
promise.
\71\Lisa Kaplan, The Biggest Social Media Operation You've Never
Heard of Is Run Out of Cyprus by Russians, Lawfare, (Dec. 18, 2019),
https://www.lawfareblog.com/biggest-social-media-operation-youve-never-
heard-run-out-cyprus-russians.
\72\Id.
\73\Id.
\74\Id.
---------------------------------------------------------------------------
Instagram, one of the leading repositories for memes and
owned by Facebook, actually had substantially more user
engagement with content posted by the Russian IRA than Facebook
did in 2016, despite having a smaller overall user base.\75\ A
panel of experts from Columbia University and two research
firms, New Knowledge and Canfield Research, prepared a report
for the Senate Intelligence Committee in which they concluded
that there were 187 million user engagements with IRA material
on Instagram--more than twice as many as on Facebook (77
million) or Twitter (73 million).\76\ It is possible that much
of the disinformation campaigns will move to Instagram to take
advantage of the younger audience.\77\
---------------------------------------------------------------------------
\75\Renee DiResta et al., New Knowledge, The Tactics and Tropes of
the Internet Research Agency, New Knowledge at 9 (2018) (upon request
from the U.S. Senate Select Committee on Intelligence), https://
int.nyt.com/data/documenthelper/533-read-report-internet-
research-agency/7871ea6d5b7bedafbf19/optimized/full.pdf.
\76\Id. at 7, 32; see also S. Rep. No. 116-XX, Volume 2 at 48-50
(2019).
\77\Taylor Lorenz, Instagram Is the Internet's New Home for Hate,
The Atlantic (March 21, 2019) https://www.theatlantic.com/technology/
archive/2019/03/instagram-is-the-internets-new-home-for-hate/585382/.
---------------------------------------------------------------------------
Instagram is poised to play a significant role in the 2020
election and beyond due to the popularity of sharing visual
content and engagement on social issues, especially among young
Americans, the ease of sharing content on the platform, and the
greater challenges in identifying corrupt activity on its
feeds.\78\ Instagram is owned by Facebook, so it has been able
to leverage the power of Facebook's vast resources, including
data and capital.\79\ Furthermore, due to Instagram's picture
based sharing format, it is rapidly becoming the platform of
choice for those who wish to peddle misinformation and false
news stories in an easily digestible and rapidly dispersed
manner.\80\ Importantly, the spread of false information and
proliferation of spoofed accounts is more complicated to detect
because of Instagram's visual medium as opposed to Facebook or
Twitter, where text-based content is predominately shared.\81\
---------------------------------------------------------------------------
\78\Allan Smith, Facebook's Instagram Poised to Be 2020
Disinformation Battleground, Experts Say, NBC News (Oct. 21, 2019),
https://www.nbcnews.com/tech/tech-news/facebook-s-instagram-poised-be-
2020-disinformation-battleground-experts-say-n1063941.
\79\Id.
\80\Id.
\81\Id.
---------------------------------------------------------------------------
Text based posts can be analyzed by automated systems to
detect origination and identify malign posts very efficiently
by the platforms.\82\ Memes spread on Instagram pose a
different and specific danger because they require additional
human review to make nuanced determinations as to whether they
are being shared as parody, satire, and other forms of humor or
if the meme is intentionally spreading misinformation and
originated with a malicious actor, such as the Russian IRA.\83\
Facebook notes that its enforcement is based on behavioral
patterns, so whether someone is sharing a meme or text, the
deceptive patterns behind that behavior (such as fake accounts,
coordinated infrastructure, etc.) will still be identifiable.
However, disinformation can often be spread without inauthentic
behavior, such as when an unsuspecting user views and spreads a
meme believing it to be real or not knowing that it originated
from a malicious actor. ``Campaigns begin with posts in blogs
or other news outlets with low standards. If all goes well,
somebody notable will inadvertently spread the disinformation
by tweet, which then leads to coverage in bigger and more
reputable outlets. The problem is, taking the trouble to
correct disinformation campaigns like these can unintentionally
satisfy the goal of spreading the meme as far as possible--a
process called amplification. Memes online make hoaxes and
psychological operations easy to pull off on an international
scale.''\84\ In effect, efforts to correct disinformation or
provide necessary factual context for misleading news may
actually result in drawing greater attention and more views to
the original disinformation.
---------------------------------------------------------------------------
\82\HVAC Committee Hearing Transcript, at 49.
\83\HVAC Interview with Nathaniel Gleicher on Nov. 1, 2019.
\84\Joan Donovan, How Memes Got Weaponized: A Short History, MIT
Technology Review (Oct. 24, 2019), https://www.technologyreview.com/
2019/10/24/132228/political-war-memes-disinformation/.
---------------------------------------------------------------------------
According to disinformation analysts, viral memes and
videos are very popular among perpetrators due to their
virtually untraceable origins, ease of construction, and rapid
dissemination to a wide audience.\85\ Another reason these
types of memes spread so efficiently on Instagram is account
administrators make their pages private which in turn requires
users to subscribe in order to view the content--this tactic
increases subscribers and results in more users seeing the
pages' posts on their feeds as opposed to someone sending it to
them through private messaging.\86\
---------------------------------------------------------------------------
\85\Allan Smith, Facebook's Instagram Poised to Be 2020
Disinformation Battleground, Experts Say, NBC News (Oct. 21, 2019),
https://www.nbcnews.com/tech/tech-news/facebook-s-instagram-poised-be-
2020-disinformation-battleground-experts-say-n1063941.
\86\Taylor Lorenz, Instagram is the Internet's New Home for Hate,
The Atlantic (Mar. 21, 2019), https://www.theatlantic.com/technology/
archive/2019/03/instagram-is-the-internets-new-home-for-hate/585382/.
---------------------------------------------------------------------------
Facebook and Twitter have taken starkly divergent
approaches to regulating certain political content on their
respective platforms. While Twitter no longer allows any paid
political advertising,\87\ Facebook continues to allow paid
advertising. Moreover, Facebook's policy is that speech and
opinions from politicians (elected officials, candidates, or
their campaigns) are not eligible to be fact-checked.\88\
Facebook has engaged independent third-party fact-checking
partners to whom it delegates the verification
responsibilities, including considerable discretion in
selecting content to be reviewed.\89\ Although Facebook claims
that advertisements, including political advertisements, on the
platform are ``eligible'' to be fact-checked by its third-party
partners, the broad exclusions for political statements and
opinions seem to effectively nullify the potential benefits of
any such verification. Facebook notes that advertisements from
Super PACs or other outside groups will still be subject to
fact checking, and that if a politician shares another user's
post that has already been fact-checked, the politician's post
will show the same warning labels from fact-checkers.\90\
---------------------------------------------------------------------------
\87\Twitter, Political Content Policy, https://
business.twitter.com/en/help/ads-policies/ads-
content-policies/political-content.html.
\88\Facebook, Fact-Checking Program Policies, https://
www.facebook.com/business/help/315131736305613?id=673052479947730; see
also Mike Isaac and Cecilia Kang, Facebook Says It Won't Back Down From
Allowing Lies in Political Ads, The New York Times (January 2020),
https://www.nytimes.com/2020/01/09/technology/facebook-political-ads-
lies.html.
\89\Facebook, Fact-Checking on Facebook, https://www.facebook.com/
business/help/2593586717571940?id=673052479947730.
\90\Facebook, Fact-Checking Program Policies, https://
www.facebook.com/business/help/315131736305613?id=673052479947730.
---------------------------------------------------------------------------
On its face, Facebook's policy would seem to create a ripe
opportunity to post disinformation in the guise of
advertisements containing ``political opinion,'' and creates a
loophole to avoid fact checking or verification. When any such
advertising campaign is specifically oriented towards veterans
or veterans' issues, an imminent risk arises of directly
channeling disinformation to veterans and VSOs. Moreover,
neither policy addresses the distribution of propaganda,
political disinformation, or doctored media for political
purposes when such distribution occurs outside of the context
of official paid political advertising. Facebook does label
content from state-controlled media entities to enable users to
identify news articles posted by these official channels.\91\
Ahead of the 2020 election, Facebook is blocking ads from
state-controlled media outlets targeted to people in the
US.\92\
---------------------------------------------------------------------------
\91\Nathaniel Gleicher, Labeling State-Controlled Media On
Facebook, Facebook: Blog (June 4, 2020), https://about.fb.com/news/
2020/06/labeling-state-controlled-media/ (last updated August 31,
2020).
\92\Id.
---------------------------------------------------------------------------
The Facebook political advertising loophole has already
been exploited to distribute some controversial content. For
example, an advertisement by the Trump campaign alleging that
Vice President Biden withheld $1 billion in U.S. foreign aid to
Ukraine to pressure the country into removing a prosecutor
investigating a firm affiliated with Biden's son was posted and
allowed on Facebook, but other outlets rejected or removed the
ad for unsubstantiated or misleading claims.\93\ Facebook
specifically cited its policy on politicians and campaigns in
its response to the Biden campaign, rejecting the request for
removal.\94\ Twitter also allowed this ad, although it came
before its ban on paid political advertising.\95\
Factcheck.org, one of the leading nonprofit arbiters of truth
and deception in politics, determined that the ad was
misleading.\96\ More recently, the Trump campaign has itself
asked Facebook to remove a video from Vice President Biden's
account that contains quotations from an Atlantic article which
purport to show President Trump repeatedly disparaging veterans
and the military.\97\ The Trump campaign notes that President
Trump denies all of the quotations attributed to him, and
therefore the video should be considered false and
misleading.\98\ However, several news organizations have
independently verified parts of the disputed allegations and
maintain the accuracy of the claims.\99\
---------------------------------------------------------------------------
\93\Cecilia Kang, Facebook's Hands-Off Approach to Political Speech
Gets Impeachment Test, The New York Times (Oct. 14, 2019), https://
www.nytimes.com/2019/10/08/technology/facebook-trump-biden-ad.html.
\94\Id.
\95\Emily Stewart, Facebook is refusing to take down a Trump ad
making false claims about Joe Biden, Vox (Oct. 9, 2019), https://
www.vox.com/policy-and-politics/2019/10/9/20906612/trump-campaign-ad-
joe-biden-ukraine-facebook.
\96\Eugene Kiely & Robert Farley, Fact: Trump TV Ad Misleads on
Biden and Ukraine, FactCheck.org (Oct. 9, 2019), https://
www.factcheck.org/2019/10/fact-trump-tv-ad-misleads-on-biden-and-
ukraine/.
\97\Paul Bedard, Outraged Trump Demands Biden, Twitter, and
Facebook Pull Down Troop Ad, Washington Examiner (Sept. 10, 2020),
https://www.washingtonexaminer.com/washington-
secrets/outraged-trump-demands-biden-twitter-facebook-pull-down-troop-
ad; see also Jeffrey Goldberg, Trump: Americans Who Died in War Are
`Losers' and `Suckers', The Atlantic (Sept. 3, 2020), https://
www.theatlantic.com/politics/archive/2020/09/trump-americans-who-died-
at-war-are-losers-and-suckers/615997/.
\98\Id.
\99\See Colby Itkowitz, Alex Horton, & Carol D. Leonnig, Trump Said
U.S. Soldiers Injured and Killed in War Were `Losers,' Magazine
Reports, The Washington Post (Sept. 4, 2020), https://
www.washingtonpost.com/politics/trump-said-us-soldiers-injured-and-
killed-in-war-were-losers-magazine-reports/2020/09/03/6e1725cc-ee35-
11ea-99a1-71343d03bc29_story.html; James LaPorta, Report: Trump
Disparaged US War Dead as `Losers,' `Suckers', The Associated Press
(Sept. 3, 2020), https://apnews.com/b823f2c285641a4a09a96a0b195636ed;
see also, Peter Baker & Maggie Haberman, Trump Faces Uproar Over
Reported Remarks Disparaging Fallen Soldiers, The New York Times (Sept.
4, 2020), https://www.nytimes.com/2020/09/04/us/politics/trump-
veterans-losers.html; Alex Ward, Did Trump Call US War Dead ``Losers''
and ``Suckers''? The Controversy, Explained., Vox (Sept. 4, 2020),
https://www.vox.com/2020/9/4/21422733/atlantic-trump-military-suckers-
losers-explained.
---------------------------------------------------------------------------
Commercial Fraud & Scams
Veterans themselves can also be direct victims of spoofing
in cases of commercial fraud. Imposters use fake social media
accounts, often posing as a VSO or other veteran interest
group, to defraud the victim by selling fake merchandise,
obtaining financial data, or even illegal fundraising.\100\ A
2017 report prepared by the American Association of Retired
Persons (AARP), found that ``more than twice as many veterans
as nonveterans lost money to scam artists during the past five
years. Some of the scams were aimed specifically at programs
and charities geared to veterans.''\101\
---------------------------------------------------------------------------
\100\Goldsmith, VVA Report.
\101\David Frank, Veterans Twice as Likely to Be Scammed, AARP:
Scams & Fraud (Nov. 8, 2017), https://www.aarp.org/money/scams-fraud/
info-2017/veterans-scam-protection-fd.html.
---------------------------------------------------------------------------
Commercial fraud aimed at veterans plays on many of the
same, emotionally-triggering themes as used in the political
propaganda campaigns, but instead of pursuing endorsement and
distribution of specific content, these scams involve financial
transactions. At one end of the scale, the scam is a simple
one-time fraudulent purchase (for example, unsanctioned
memorabilia, or fake/ knock-off merchandise). The more
sophisticated and devious plots aim to extract larger sums of
money over longer time periods, or in the extreme example, even
obtain the victim's actual financial information.
An important subset of the online fraud perpetrated against
or through veterans is the category of romance scams, in which
scammers pose as veterans seeking a relationship and send
requests to victims for money based on fabrications. Spoofers
appropriate real veterans' images and stories, including
veteran families' grief and hardships--in order to scam
individuals who are sympathetic and supportive of veterans. The
overall volume of online fraud claims runs into the billions of
dollars and is increasing.\102\
---------------------------------------------------------------------------
\102\Federal Bureau of Investigation, U.S. Dep't of Justice, 2019
Internet Crimes Report (on file at https://pdf.ic3.gov/
2019_IC3Report.pdf).
---------------------------------------------------------------------------
In instances of financial fraud or romance scams, criminals
are exploiting the general sense of trust that the American
people have in those who serve in uniform. People lower their
guard when interacting with someone who is serving the country,
and that includes when interacting online. There is a large
organized crime ring based in Nigeria that recognizes this and
has built an industry around stealing veterans' identities for
use in financial scams. These men in Nigeria proudly call
themselves ``Yahoo Boys,'' a nickname that came about in the
1990's based on email scams from supposed ``Nigerian Princes''
who offered huge deposits in exchange for private banking
information.\103\
---------------------------------------------------------------------------
\103\Jack Nicas, Facebook Connected Her to a Tattooed Soldier in
Iraq. Or So She Thought, The New York Times (July 28, 2019), https://
www.nytimes.com/2019/07/28/technology/facebook-
military-scam.html.
---------------------------------------------------------------------------
Online criminals often steal veterans' deployment photos
and use them to create online social media profiles. They then
use those imposter profiles to enter online groups which are
made for grieving Gold Star families. These predators know that
with a military death comes a large life insurance payout, so
they use stolen identities to comfort widows and widowers,
offering love and attention. After weeks or months of grooming
a victim, forming what the victim believes to be a romantic
relationship, the scammers will make up stories about being in
desperate financial situations. Victims will often send large
sums of money believing that they are helping a service member
in need, or to pay for an airline ticket for facilitating a
romantic meeting. Then the scammers doctor photos of plane
tickets and send them to victims. Victims often end up waiting
at an airport for hours before they come to realize the
scam.\104\
---------------------------------------------------------------------------
\104\Id.
---------------------------------------------------------------------------
News reports have documented several cases where victims of
these scams die by suicide after realizing that they were
tricked into giving away their life savings.\105\ The subject
of a New York Times article on veteran-based romance scams, one
individual lost between $26,000 and $30,000 in just two years
to an imposter posing as a veteran.\106\ After sending the
imposter $5,000 for what was supposed to be plane tickets to
visit, the victim attempted suicide.\107\ During the
investigation for the New York Times article, this spoofing
victim was killed by her husband, who also killed himself and
the victim's father.\108\
---------------------------------------------------------------------------
\105\Id.
\106\Id.
\107\Id.
\108\Id.
---------------------------------------------------------------------------
WHAT SPOOFING LOOKS LIKE
The effectiveness of spoofing campaigns lies in the ability
of the spoofer to present content online in a manner that
appears ordinary and credible, while actually advancing a
malicious intent. The examples below document how spoofing
manifests in both the spreading of political propaganda and the
perpetration of commercial fraud.
Political Propaganda & Socially Divisive Content
The image below on the left was posted by the Facebook
group ``Vietnam Veterans'' in January 2020.\109\ ``Vietnam
Veterans'' has stolen content from a nationally chartered VSO
and has ties to pages known to be operated from outside the
U.S.\110\ The image depicts former professional football player
Pat Tillman, who quit the National Football League and enlisted
as an Army Ranger in response to the terrorist attacks of
September 11, 2001. Tillman was subsequently killed in action.
The image of Tillman is juxtaposed with a caption disparaging
Colin Kaepernick, another former National Football League (NFL)
player who gained notoriety for his pre-game protests in which
he knelt during the National Anthem to draw attention to the
issues of police brutality and racial disparities in police
shootings. The image on the right features Kaepernick again,
this time contrasting him with Glen Coffee, another former NFL
player who enlisted in the Army.\111\
---------------------------------------------------------------------------
\109\Vietnam Veterans Facebook Page--a page purportedly for and run
by veterans, is a spoofing page that drives people to merchandise sites
and is run by zero individuals in the United States (an indicator of a
spoofed page). Link
\110\Goldsmith, VVA Report.
\111\These images can be found here: Goldsmith, VVA Report, https:/
/vva.org/wp-content/uploads/2019/09/VVA-Investigation.pdf.
In both instances, Kaepernick is being contrasted with
other former NFL players who left professional football to join
the military, ostensibly to differentiate the privileged
athlete from those who sacrificed the same privilege in order
to serve the country. These are examples of socially divisive
images being used to place veterans and ``heroes'' on one side
and those protesting police brutality, or supporting Black
Lives Matter, on the other. The first image is attempting to
position veterans against Kaepernick's protest movement, which
is closely associated with liberal sentiments and especially
with racial minorities. By delineating the groups this way,
this image also aligns veterans with law enforcement, further
emphasizing that one side represents heroes, while the other
side represents liberals, ``sissy's'' [SIC], and perhaps
minorities. The second image comes from a page called ``Vets
For Trump'' that makes the same distinction, but with a more
overt partisan affiliation.
Images that focus on divisive issues that fall on political
fault lines are used to drive interactions for many purposes--
commercial fraud, misinformation, and romance scams among many
others. The following images highlight some of the political
pressure points that spoofers use to increase the number of
users exposed to their schemes. The image below, also posted by
``Vietnam Veterans,'' references the same Kaepernick protest
with the text above the picture calling out ``overpaid
kneelers'' and is meant to leverage the pain and loss felt by
military families. Further, by conflating the issues and
sacrifices of military members and veterans with issues of race
and police enforcement, bad actors are able to sow anger and
division. By artificially positioning these groups of
``heroes'' as opposed to everyone else, spoofers manipulate an
emotional response and then call for ``sharing'' the post,
leading to significantly increased exposure.\112\
---------------------------------------------------------------------------
\112\Id.
The next image was posted by ``Vets for Trump'' and
attempts to create the illusion that President Obama did not
care about the military, while representing President Trump as
someone who will take care of the troops. This type of imagery
is misleading and pits two segments of the population against
one another, Democrats and Republicans, and paints one party as
respecting the military and the other as disrespecting it. The
group ``Vets for Trump'' was run by individuals outside the
United States at one point, and fit the profile of activity
outlined in the Senate Select Committee on Intelligence report
on foreign interference.\113\ Facebook restored the Vets for
Trump Page to its original owners in August 2019.\114\
---------------------------------------------------------------------------
\113\S. Rep. No. 116-XX, Volume 2 at 8 (2019).
\114\Goldsmith, VVA Report, at 145.
This image below creates a false choice between veterans'
issues and immigration issues. By conflating the two issues,
foreign actors are able take a point of agreement--veterans'
issues--and turn it into a pressure point of partisan fighting.
Pushing this type of content drives page engagements and
establishes a user base on whom spoofers can later run
commercial or romance schemes, in addition to creating
political interference. The image was posted by ``Veterans
Nation'' which is run by a collection of administrators from
Vietnam, Brazil, and Ukraine--notably none of the
administrators are based in the United States. Furthermore, the
``Veterans Nation'' group shares the same content created by
the ``Vets for Trump'' page.\115\
---------------------------------------------------------------------------
\115\These images can be found here: Goldsmith, VVA Report, https:/
/vva.org/wp-content/uploads/2019/09/VVA-Investigation.pdf.
Divisive issue pushing is not unique to any one political
group or viewpoint. Below is a screen grab from ``Vietnam
Veterans Advocacy Group,'' which shared only pro-Obama and
anti-Trump articles from unreliable and questionable websites.
The article attempts to smear President Trump using rumors and
tabloid-style headlines. By driving views from both sides of
the political divide, foreign influence operations can
effectuate the sowing of discord and distrust in American
institutions highlighted in the Senate Select Committee on
Intelligence Report.\116\ The capitalization on political
animosity is a driving force for misinformation, as well as the
other fraudulent schemes spoofers attempt to execute.
---------------------------------------------------------------------------
\116\S. Rep. No. 116-XX, Volume 2 at 8 (2019).
Commercial Fraud
After driving user interactions with spoofed pages, the
schemes often turn their efforts towards gaining profit through
the sale of counterfeit products. The counterfeit products
range from coins to flags and often use stolen intellectual
property or copyright insignia. The images below show examples
of commercial fraud, including the sale of products such as
counterfeit commemorative coins, knives, and flags, often using
stolen intellectual property. The first image shows the spoofed
site ``Vietnam Vets Unite'' Facebook page linking to an Amazon
store. Once a user selects the link, the user is redirected to
an Amazon store offering counterfeit VVA-branded merchandise
from the seller Coins and Coins. The second image shows the
Amazon storefront, with the counterfeit VVA-branded coin
images. This scheme is repeated across different Facebook
groups and with a multitude of counterfeit items.\117\
---------------------------------------------------------------------------
\117\Goldsmith, VVA Report.
A SPOOFING CASE STUDY--VIETNAM VETERANS OF AMERICA
The Committee reviewed an in-depth investigative report on
internet spoofing specifically targeting veterans that was
researched and prepared by Kristofer Goldsmith of VVA.\118\ Mr.
Goldsmith also appeared before the Committee to offer testimony
about his research.\119\
---------------------------------------------------------------------------
\118\Id.
\119\HVAC Committee Hearing Transcript, at 12.
---------------------------------------------------------------------------
In August of 2017, VVA discovered an imposter Facebook page
that misappropriated VVA's trademarked name and logo and was
linked to a suspicious European-based website. The spoofed page
was spreading falsified news stories on issues specifically
associated with veterans. The discovery of the fake page led
VVA to begin an investigation into online spoofing, which
ultimately revealed a history of ``persistent and pervasive
foreign-born online campaigns'' that had targeted the group and
its members since at least 2014.\120\ After a few months of
investigation, VVA shared its findings with law enforcement
agencies and congressional committees, including this
Committee. The initial findings identified an entity in
Plovdiv, Bulgaria, as creator and manager of the spoofed
page.\121\
---------------------------------------------------------------------------
\120\Goldsmith, VVA Report.
\121\Id.
---------------------------------------------------------------------------
VVA eventually recognized that this instance of spoofing
actually represented a more pervasive problem, stating:
American veterans and the social-media followers of
several congressionally chartered veterans service
organizations were specifically targeted by the Russian
Internet Research Agency with at least 113 ads during
and after the 2016 election. However, this represents
but a tiny fraction of the Russian activity that
targeted this community with divisive propaganda: The
organic politically divisive content (organic meaning
not having to do with ads, rather unpaid posts and
comments) created by Russians have a far greater reach
than the known paid ads; for even though many of the
original sources have been removed from social-media
platforms, their posts and comments continue to be
propagated and disseminated by foreign administrators
(aka admins, who maintain and manage online sites) to
spread hateful and politically divisive messages.\122\
---------------------------------------------------------------------------
\122\Id. at 6.
---------------------------------------------------------------------------
VVA Encounters Challenges to Take Down Spoofed Site
After the discovery of the spoofed page on August 21, 2017,
VVA contacted a member of the Facebook security team on August
23, 2017, to notify them of the unauthorized and
misappropriated use of VVA's name and logo, and to request that
the fraudulent page be taken down. Following this initial
notification, the spoofed page remained active and on September
26, 2017, the site shared a manipulated video that resulted in
over 37,000 views by October 3, 2017. VVA again reported the
page to Facebook. A week later, on October 9, 2017, with
Facebook offering no solution, VVA went public with appeals to
the Department of Defense (DoD) and the Department of Veterans
Affairs (VA), requesting measures to protect service members
and veterans from online foreign political influence.\123\ By
mid-October of 2017, Facebook stated that the spoofed page had
not violated terms of service and placed the burden of
clarification back on VVA.\124\ On October 24, 2017, Facebook
finally removed the spoofed page, but only due to a finding
that the page had violated VVA's copyright.\125\ To date,
neither DoD nor VA have responded to VVA's request for measures
to protect service members and veterans.
---------------------------------------------------------------------------
\123\Leo Shane, Report: Online Trolls Targeting US Troops,
Veterans, Military Times (Oct. 10, 2017), https://
www.militarytimes.com/veterans/2017/10/10/report-online-trolls-
targeting-us-troops-veterans/.
\124\Nikki Wentling, Veterans Organization Asks for More Help
Combating `Imposter' Facebook Page, Stars and Stripes (Oct. 18, 2017),
https://www.stripes.com/news/veterans-organization-asks-for-more-help-
combating-imposter-facebook-page-1.493168.
\125\Id.
---------------------------------------------------------------------------
Growth of Spoofed Site
VVA found it very challenging to convince Facebook to take
down a spoof of its legitimate Facebook page, but the spoofed
page was eventually taken down by Facebook. Facebook cited the
copyright issues posed by the spoofed page as the reason for
the page being removed and not spoofing.\126\ However, in the
interim, the incredibly rapid growth of the spoofed page made
it difficult for users to recognize the spoofed VVA Facebook
page as a spoof. The time from first notification to Facebook
to the removal of the page was approximately two months. During
that time the spoofed page gained nearly 200,000 followers and
significantly more impressions.\127\
---------------------------------------------------------------------------
\126\Goldsmith, VVA Report, at 31.
\127\Id. at 25.
---------------------------------------------------------------------------
As the VVA report explains, a large number of followers
provides credibility for a spoofed page. Spoofers can increase
the number of followers exponentially by distributing posts
that are a mix of politically divisive (such as memes of
politicians and policy agendas) and generally soothing posts
(such as crafting or animal videos). This mixture of content
drives the number of likes, shares, comments, and interactions
which in turn escalate the influence of these malign actors.
Spoofed accounts or pages frequently feature a pattern of
rapid growth of the subscriber base, which often massively
surpasses the subscriber base of legitimate veterans' and VSO
pages. Very rapid growth for a new site, and particularly for
sites whose posting patterns are irregular, voluminous, or
repetitive, may indicate spoofing activity. VVA noted, ``The
rate at which the [fake] `Vietnam Vets of America' page grew in
followers is staggering. According to their `About' tab, they
went from 30,000 followers on November 1, 2016, to 196,567 as
of October 2017. For comparison, the real VVA page has only
garnered approximately 137,000 likes since it was created in
June 2010.''\128\
---------------------------------------------------------------------------
\128\Id.
---------------------------------------------------------------------------
Conclusion of VVA Investigation
After discovering the spoofed site, conducting an initial
investigation, and notifying Facebook, Congress, and VA, VVA
then took the initiative to conduct a full, two-year
investigation on spoofing and the veterans community. The
investigation resulted in the documentation of ``persistent,
pervasive, and coordinated online targeting of American
servicemembers, veterans, and their families by foreign
entities who seek to disrupt American democracy.''\129\ During
the Committee hearing, Rep. Michael Bost (R-IL) queried Mr.
Goldsmith about what VVA was doing to inform and assist
veterans with the problems caused by spoofing.\130\ Mr.
Goldsmith noted that VVA primarily uses Facebook and Twitter to
educate and communicate with veterans. So, when spoofers use
Facebook and Twitter to spread disinformation it becomes very
difficult for veterans to distinguish legitimate content from
illegitimate content.\131\ Mr. Goldsmith went on to say that
this problem illustrates the urgent need for the social media
platforms and the VSOs to develop strategies to help veterans
identify potential disinformation online.\132\
---------------------------------------------------------------------------
\129\Id. at 6.
\130\HVAC Committee Hearing Transcript, at 65.
\131\Id.
\132\Hijacking Our Heroes: Exploiting Veterans through
Disinformation on Social Media Before the H. Comm. On Veterans'
Affairs, 116th Cong. (2019) (written testimony of Mr. Kristofer
Goldsmith, Chief Investigator & Associate Dir. Of Policy & Gov.
Affairs, Vietnam Veterans of America, at 11-12) (access Mr. Goldsmith's
written testimony here, https://docs.house.gov/meetings/VR/VR00/
20191113/110183/HHRG-116-VR00-Wstate-GoldsmithK-20191113.pdf).
---------------------------------------------------------------------------
Based on its own experience of being spoofed, and
considering the lessons gleaned from its extensive
investigation, VVA recommended that the social media platforms
draw upon the resources within the veterans' community by
partnering with the VSOs in order to help raise awareness of
the problems and permutations of spoofing. Additionally, such
partnerships would also provide the social media platforms with
access to military and veteran expertise that could help refine
the platforms' ability to detect and discern misrepresentation
or fraud targeted at veterans. Finally, VVA also urged all
parties to collaborate in facilitating assistance to victims of
spoofing by streamlining and expediting the process of
reporting and removing spoofing activity on the platforms.
Scope of the Spoofing Problem
In trying to determine the scope of the problem posed by
internet spoofing, the Committee requested an analysis by
Graphika, a firm specializing in the study of social networks,
data manipulation, and how messaging on these networks evolves
and spreads.\133\ Dr. Vlad Barash of Graphika performed a study
of disinformation campaigns targeting American veterans and
military service members to understand the volume and timeframe
of these messaging campaigns, as well as specific details of
the targeted communities and the substantive message contents.
---------------------------------------------------------------------------
\133\Graphika, The Graphika Platform, https://graphika.com/how-it-
works.
---------------------------------------------------------------------------
Graphika based its analysis on one dataset collected for a
previous study, several datasets that were publicly released by
Twitter following discovery and verification of state-backed
foreign ownership, and one Facebook dataset that was developed
and collected by VVA.\134\ Graphika determined that just 2,106
Twitter accounts associated with veterans and/or military
personnel were able to ultimately reach over 5,000,000 Twitter
accounts. Similarly, on Facebook, Graphika found that a mere 41
pages oriented at veterans or service members reached a total
of 18,298,968 followers. Both results revealed a ``powerful
multiplier effect'' that extended the reach and potential
audience achievable through the manipulation of a relatively
small number of social media pages or accounts.\135\ Moreover,
Graphika identified a troubling trend in its analysis of
disinformation operations. The rate of activity targeting
American veterans and military service members has increased,
not decreased, since the 2016 U.S. election.\136\
---------------------------------------------------------------------------
\134\Graphika answered these questions by analyzing three types of
datasets: an initial study of foreign operations targeting US Veterans
by the Oxford Internet Institute and Graphika (Gallacher et al., Junk
News on Military Affairs and National Security: Social Media
Disinformation Campaigns Against US Military Personnel and Veterans
(2017), http://comprop.oii.ox.ac.uk/research/working-papers/vetops/);
datasets of foreign information operations on Twitter, curated and
publicly released by the company (Vijaya Gadde, & Yoel Roth, Enabling
further research of information operations on Twitter, Twitter: Company
Blog (Oct. 17, 2018), https://blog.twitter.com/en_us/topics/company/
2018/enabling-further-research-of-information-
operations-on-twitter.html); and a dataset collected by Kristofer
Goldsmith of VVA of activity around Facebook public pages on veteran
and military-related topics with foreign administrator accounts
(Kristofer Goldsmith, Vietnam Veterans of America, An Investigation
Into Foreign Entities Who Are Targeting Servicemembers and Veterans
Online (2019), https://vva.org/wp-content/uploads/2019/09/VVA-
Investigation.pdf). See Hijacking Our Heroes: Exploiting Veterans
through Disinformation on Social Media Before the H. Comm. On Veterans'
Affairs, 116th Cong. (2019) (written testimony of Dr. Vlad Barash,
Science Dir., Graphika) (access Dr. Barash's written testimony here,
https://docs.house.gov/meetings/VR/VR00/20191113/110183/HHRG-116-VR00-
Wstate-BarashV-20191113.pdf).
\135\Dr. Barash written testimony, at 7, n.25.
\136\Id., at 4, n.15.
---------------------------------------------------------------------------
Graphika continues to uncover and expose ongoing
information operations that target the 2020 Presidential
election, making these types of campaigns a persistent threat
for our democracy. Graphika anticipates that these campaigns
will continue to target influential American communities,
including veterans and the military.\137\ Previous research has
found that ``U.S. veterans and members of our military are
highly respected members of society who ``positively influence
their country and their community.''\138\ Graphika's analysis
of the 2,106 veteran-associated Twitter accounts mentioned
above identified some of them as ``influencers'' in this
discussion.\139\ ``Influencers'' are individual accounts that
have a disproportionate impact in trend setting and
distribution of content, and are exceptionally valuable for
marketers, entertainers, fashion/consumer goods labels and in
the case of those seeking to spread disinformation or political
propaganda, have the ability to reach a broad audience very
quickly and efficiently.\140\ The specific targeting of
influencers to quickly and broadly disseminate messages is a
very effective and dangerous tactic. A key influencer
retweeting or posting about a single piece of disinformation
can significantly amplify the impact and reach of that
disinformation, especially as compared to a non-influencer.
---------------------------------------------------------------------------
\137\HVAC Majority Staff Interview with Dr. Vlad Barash on November
5, 2019.
\138\Drew Lieberman & Kathryn Stewart, Greenberg Quinlan Rosner
Research, Strengthening Perceptions of America's Post-9/11 Veterans:
Survey Analysis Report (2014), https://www.dillonconsult.com/wp-
content/uploads/2013/03/Strengthening-Perceptions-of-Americas-Post-911-
Veterans-Survey-Analysis-Report-Got-Your-6-June-2014.pdf.
\139\Gallacher et al., Computational Propaganda Research Project,
Junk News on Military Affairs and National Security: Social Media
Disinformation Campaigns Against US Military Personnel and Veterans
(2017), http://comprop.oii.ox.ac.uk/research/working-papers/vetops/.
\140\Ismael El Qudsi, What To Expect For Influencer Marketing in
2020, Forbes (Feb. 6, 2020), https://www.forbes.com/sites/
forbesagencycouncil/2020/02/06/what-to-expect-for-influencer-
marketing-in-2020/#491d7f965c09.
---------------------------------------------------------------------------
Graphika also noted that the Twitter posts examined in its
study generally referenced key topics of particular interest to
U.S. service members or veterans. This included messages that
were positive, such as supporting troops, or negative, such as
discussing the challenges of post-traumatic stress disorder or
homelessness among the veteran community.\141\ As discussed
above, the use of carefully selected topics to trigger an
emotional response along with an endorsing action (liking,
sharing, or retweeting) is a common technique used by spoofers
to quickly disseminate their content with the imprimatur of an
authoritative voice.\142\ Foreign based spoofers are then able
to inject their own agenda and propaganda into the discussion
around these important subjects, without the knowledge of the
readers and viewers who receive the content from an ostensibly
authentic and authoritative source.\143\
---------------------------------------------------------------------------
\141\Dr. Barash written testimony, at 2.
\142\Goldsmith, VVA Report, at 25.
\143\Id.
---------------------------------------------------------------------------
Graphika has observed the effectiveness of Russian and
Iranian operations in targeting American audiences with
disinformation in order to sow public discord.\144\ Foreign
information operations targeting U.S. veterans and military
members are found across social media platforms, have been
ongoing since at least 2011 and are steadily growing, according
to Graphika's analysis.\145\ Although when taken in the context
of the overall scale of all social media content, these
operations only account for a very small number of messages or
pages, the volume of the raw data belies its impact.
Additionally, the content of the messages demonstrates that
they were carefully crafted to hijack key topics of discussion
among U.S. veterans and military service members, for example
by inserting calls to violence into positive messages around
troop support.''\146\
---------------------------------------------------------------------------
\144\HVAC Committee Hearing Transcript, at 44.
\145\Dr. Barash written testimony, at 3.
\146\Id. at 9.
---------------------------------------------------------------------------
Ranking Member Dr. Roe asked Dr. Barash, ``First, are
veterans targeted for scams at a higher rate than non-
veterans'' and ``. . . secondly, are veterans targeted for
propaganda at a higher rate than non-veterans'' (emphasis
added).\147\ Dr. Barash responded, ``Yes and yes. Veterans are
an influential community in our social fabric online and
offline. And as a result, it is much more effective to target
them with all kinds of operations including propaganda.''\148\
---------------------------------------------------------------------------
\147\HVAC Committee Hearing transcript at 44.
\148\Id.
---------------------------------------------------------------------------
Graphika's ultimate conclusions about Twitter both support
and complement VVA's findings regarding Facebook. Graphika's
analysis demonstrates that the contents of the data sets
indicated a precisely targeted campaign to exploit an
influential American community in order to spread
disinformation as broadly and as persuasively as possible and
not randomly generated Tweets.\149\ This result mirrors the
Facebook example documented by VVA, where just a few foreign-
run pages oriented at veterans successfully reached an audience
of millions.\150\
---------------------------------------------------------------------------
\149\Dr. Barash written testimony, at 4.
\150\Goldsmith, VVA Report.
---------------------------------------------------------------------------
Dr. Barash also informed the Committee that there were
significant constraints imposed upon his analysis by the
limitations on the data sets made available by the social media
platforms. Consequently, there are still considerable barriers
to fully documenting the nature and scale of the problem. He
noted that, ``The data available so far allow for a piecemeal
analysis approach to a multi-faceted operation.''\151\ Twitter
separately confirmed to the Committee that its internal
analysis supports a finding of additional social media activity
on other platforms involving the same foreign-based accounts in
these datasets, but metrics on volume, timeframe, or content
were not available.\152\ Dr. Barash strongly stressed the need
for comprehensive data collection by the social media
platforms, and collaborative analysis based on shared access to
the data in order to make final determinations about the scope,
impact, and likely developments in information operations
against American veterans and service members.\153\
---------------------------------------------------------------------------
\151\HVAC Committee Hearing Transcript, at 22.
\152\HVAC Majority Staff Interview with Kevin Kane on October 17,
2019.
\153\Id.
---------------------------------------------------------------------------
Dr. Barash concluded that based on what he knows to date,
his analysis clearly demonstrates the need for a broad-based
approach to protecting and supporting the veteran and military
communities from foreign entities targeting them on social
media. Specifically, he suggested that the press and
educational institutions should provide resources and fact-
checking efforts specifically geared towards American veterans
in order to help promote awareness around these types of
foreign campaigns and the use of divisive content to drive the
growth behind fraudulent accounts. Furthermore, he testified,
research institutions can fund, and researchers can develop,
next-generation disinformation detection mechanisms which are
community-focused and tailored to help flag suspicious social
media content, as well as other new deterrence approaches. Dr.
Barash recommended that the major social media platforms should
work with Congress and the law enforcement agencies to take
coordinated actions to protect our veterans by bringing greater
transparency, easier access to data, and stronger detection
tools to the social media experience.\154\ Such coordination
and sharing of data would provide analysts such as Graphika
with broader, more accurate information with which to properly
understand the operations, scope, and activity patterns of the
networks used by spoofers from around the world. Cooperation
among the platforms in sharing data related to spoofing, fraud,
or criminal activity, would permit analysts to perceive
patterns, rather than solely examining discrete incidents and
then trying to extrapolate patterns.
---------------------------------------------------------------------------
\154\Id.
---------------------------------------------------------------------------
The Social Media Platforms
The Committee solicited testimony from two of the most
significant social media platforms: Facebook and Twitter.
FACEBOOK
Facebook is the largest social media platform, with 1.79
billion daily active users as of June 30, 2020.\155\
Additionally, it owns Instagram which has over 1 billion
monthly users, and messaging app WhatsApp, which has over 1.5
billion monthly users. Over 2 billion people use Facebook,
Instagram, WhatsApp, or Messenger every day on average.\156\
With this massive reach across multiple popular applications,
Facebook has unparalleled influence in the realm of social
media, which makes it particularly valuable for foreign
spoofers attempting to interject external agendas into American
political debates. Facebook was a significant vehicle of
Russian interference in the 2016 election, as discussed in the
Senate Intelligence Committee Report.\157\ As both Mr.
Goldsmith and Dr. Barash testified, there is a very real
prospect that this type of foreign threat will again be a
factor in the 2020 election. Further compounding the risk,
Facebook Chief Executive Officer Mark Zuckerberg announced in
late 2019 that Facebook will not be fact-checking any
advertisements bought by politicians, candidates, or political
campaigns, on the platform, arguing that private companies
should not be censoring politicians.\158\
---------------------------------------------------------------------------
\155\Facebook, Facebook Reports Second Quarter 2020 Results,
Facebook Investor Relations (July 30, 2020), https://investor.fb.com/
investor-news/press-release-details/2020/Facebook-
Reports-Second-Quarter-2020-Results/default.aspx.
\156\Mike Snider, Facebook reportedly looks to link Messenger,
WhatsApp and Instagram messaging, USA Today (Jan. 27, 2019), https://
www.usatoday.com/story/tech/talkingtech/2019/01/25/facebook-instagram-
whatsapp-linked-messaging-reportedly-works/2676662002/; see also Salman
Aslam, Instagram by the Numbers: Stats, Demographics & Fun Facts,
Omnicore (Feb. 10, 2020), https://www.omnicoreagency.com/instagram-
statistics/.
\157\S. Rep. No. 116-XX, Volume 2 at 8 (2019).
\158\Drew Harwell, Faked Pelosi Videos, Slowed to Make Her Appear
Drunk, Spread Across Social Media, The Washington Post (May 24, 2019),
https://www.washingtonpost.com/technology/2019/05/23/faked-pelosi-
videos-slowed-make-her-appear-drunk-spread-across-social-media/.
---------------------------------------------------------------------------
How Facebook is Combatting Spoofing
Facebook's Director of Security Policy, Mr. Nathaniel
Gleicher, testified before the Committee that the platform's
commitment to honest interactions for its users starts with a
set of basic policies to protect against what Facebook deems
inauthentic behavior--including misrepresentation, fraud,
deception, and spam. He stated that these policies are intended
to create a space where platform users can trust the people and
communities with which they are interacting. First, people are
required to connect on Facebook using real names. Second,
people are broadly prohibited from misrepresentation on
Facebook, including the use of fake/fraudulent accounts,
artificially boosting the apparent popularity of content (e.g.
using bots or machines to automatically generate positive
feedback for a given post), or to otherwise violate the
published Community Standards of Facebook. Users are
specifically prohibited from impersonating other persons, which
is the fundamental aspect of spoofing.\159\ Facebook policies
also prohibit users from maintaining multiple Facebook
profiles.\160\
---------------------------------------------------------------------------
\159\Facebook, Fraud and Deception Policy, https://
www.facebook.com/communitystandards/fraud_deception.
\160\Facebook, Inauthentic Behavior Policy, https://
www.facebook.com/communitystandards/inauthentic_behavior.
---------------------------------------------------------------------------
Notwithstanding these stated policies and the testimony of
Mr. Gleicher, there are significant and material deficiencies
in the implementation of these policies. Facebook frequently
gets media attention for its removals of fake accounts,
sometimes involving foreign actors or state entities.\161\
However, the very existence of these fake accounts in the first
place illustrates that Facebook's policies against creating
inauthentic accounts can be circumvented. Although Facebook
requires real identities to be used to create accounts, in fact
it is feasible for those real identities to be used to create
accounts under different and fraudulent names, as happened in
the example of VVA. In other words, while it is true that every
account must be rooted in a real identity, that identity may
not match the one being presented on Facebook. While the
Facebook policy requires that such accounts be removed and shut
down upon discovery, there are opportunities for spoofers to do
significant harm before they are discovered and ousted.
Additionally, Facebook also allows multiple pages to be
connected to an individual account. These multiple pages can be
misleading to the unsuspecting user who simply engages with a
page based on the name, picture, or logo, without assessing
whether the underlying account is actually the one it purports
to be, as again illustrated by the VVA example. Despite
Facebook's policies and efforts to verify the identities tied
to accounts, there continue to be opportunities for spoofers to
infiltrate the platform, at least until they are discovered.
That said, it is also clear that Facebook has invested
significantly to try to ameliorate this problem, and that these
investments have undoubtedly contributed to blocking many such
attempts to create fraudulent accounts, often automatically by
technology that Facebook has installed.\162\
---------------------------------------------------------------------------
\161\See, e.g., Queenie Wong, Facebook Takes Down Network of Fake
Accounts Tied to Infamous Kremlin-Linked Troll Farm, CNET (Sep. 2,
2020), https://www.cnet.com/news/facebook-says-its-catching-russian-
linked-fake-accounts-earlier/.
\162\Hijacking Our Heroes: Exploiting Veterans through
Disinformation on Social Media Before the H. Comm. On Veterans'
Affairs, 116th Cong. (2019) (written testimony of Mr. Nathaniel
Gleicher, Head of Sec. Policy, Facebook, at 4) (access Mr. Gleicher's
written testimony here, https://docs.house.gov/meetings/VR/VR00/
20191113/110183/HHRG-116-VR00-Wstate-GleicherN-20191113.pdf).
---------------------------------------------------------------------------
Additionally, Facebook also implements higher standards of
verification, visibility, and transparency for pages that
exceed a threshold for large numbers of followers, political
advertisers, and certain commercial pages.\163\ Private Groups
on Facebook, however, have emerged as a way that inauthentic
accounts attempt to work around the verification and
transparency requirements for large pages, and are therefore
increasingly becoming the distribution network of choice for
many spoofers disseminating propaganda.\164\ Although Facebook
maintains that it uses tools to detect and remove violating
content within private groups, these groups can still contain
vast networks of disinformation.\165\
---------------------------------------------------------------------------
\163\Ashley Carman, Facebook Says It'll Now Require Political-
Leaning Advertisers to Verify Their Identity, The Verge (Apr. 6, 2018),
https://www.theverge.com/2018/4/6/17206670/facebook-issue-ad-
verification-advertisers-pages.
\164\Jonathan Albright, The Shadow Organizing of Facebook Groups,
Medium (Nov. 4, 2018), https://medium.com/s/the-micro-propaganda-
machine/the-2018-facebook-midterms-part-ii-
shadow-organization-c97de1c54c65.
\165\Id.
---------------------------------------------------------------------------
Under its current spoofing enforcement structure, Facebook
features four layered lines of review. The process is first
built on automated computer detection of signals data about
account creation and usage, such as the use of suspicious email
addresses, suspicious activity patterns, or common signals
previously associated with other fake accounts that have been
removed (e.g., shared IP addresses). Facebook relies upon
technology and machine-learning review to automatically detect
and eliminate the most common threats. This reduces the noise
in the search environment for the human reviewers by removing
the most basic, crude, or unsophisticated threats, thereby
making it easier for the investigators to isolate more
sophisticated bad actors. Automated detection allows for the
rapid analysis of very large quantities of data, which enables
the detection of anomalies, discrepancies, patterns or trends
that may be indiscernible to human reviewers. As previously
noted, patterns of suspicious activity can be a more reliable
indicator of fraudulent or spoofed accounts than the more
granular review applied by human reviewers.\166\
---------------------------------------------------------------------------
\166\Mr. Gleicher written testimony, at 4.
---------------------------------------------------------------------------
The second layer is human review of actual pages, posts,
and activity. Human investigators employed by Facebook and with
experience in cybersecurity research, law enforcement, and
investigative reporting, search for and remove the most
sophisticated threats. To do so, they collaborate closely with
Facebook's data science team, which uses machine learning and
other advanced technologies to identify patterns of malicious
behavior. Human review adds a necessary and important element
that may otherwise be beyond the current abilities of machine
review, namely the subjective assessment of whether given
content violates community standards, as compared with the
objective assessment of where, when, and how that content was
posted. For example, human review can pick up on nuances and
can therefore allow permissible concepts such as parody,
satire, and privacy interests to be incorporated into the
evaluation process of taking an account down or confirming its
authenticity. Facebook has over 35,000 employees dedicated to
safety and security, including content moderation.\167\
---------------------------------------------------------------------------
\167\HVAC Committee Hearing Transcript, at 49.
---------------------------------------------------------------------------
Third, in addition to using both humans and machines to
weed out the identifiable spoofed content from the platform,
Facebook also provides users with account information so that
they can independently verify pages or affiliations. For
example, Facebook provides identity and geographic information
about certain pages, so that if a page is owned or run by a
foreign actor, the country location of the people or
organizations managing the page is easily determined and,
therefore, people can better assess whether the page they are
engaging with is legitimate and authentic. Del. Aumua Amata
Coleman Radewagen (R-AS) asked whether the platforms' takedown
and enforcement policies were at all informed by whether the
scam was perpetrated by a non-state or a state actor.\168\ Mr.
Gleicher responded that the vast majority of fraudulent
activities are committed by actors ``motivated in order to make
money'' and when working to identify state-based actors
Facebook has a number of strict controls to establish proof of
association.\169\ Facebook labels content from state-controlled
media and is blocking advertisements from such outlets ahead of
the U.S. election.\170\
---------------------------------------------------------------------------
\168\Id. at 85.
\169\Id.
\170\Nathaniel Gleicher, Labeling State-Controlled Media On
Facebook, Facebook: Blog (June 4, 2020), https://about.fb.com/news/
2020/06/labeling-state-controlled-media/ (last updated August 31,
2020).
---------------------------------------------------------------------------
According to Facebook's written testimony, users sometimes
fail to disclose organizational control of their pages in order
to make other viewers believe that the page is run
independently.\171\ Mr. Gleicher also noted that Facebook
prioritizes authentic engagement on its platform, and wants
users to understand who is speaking to them and what
perspectives are being represented.\172\ Towards this end,
Facebook has recently introduced a policy to deliver more
accountability by requiring pages that are suspected of
concealing or misrepresenting the page's ownership to go
through the formal business verification process and show more
detailed ownership information in order to remain live on the
platform.\173\
---------------------------------------------------------------------------
\171\Mr. Gleicher written testimony, at 4.
\172\Id.
\173\Id.
---------------------------------------------------------------------------
Fourth, Facebook has formed external partnerships with peer
social media platforms, law enforcement agencies, and a group
of third-party analysts (including academic researchers, think
tanks, and governments), to study removed accounts for patterns
of activity or identification data, and to more efficiently
identify emerging or potential cross-platform threats. This is
intended to create a more comprehensive understanding of
vulnerabilities and deterrence strategies that can be deployed
across the range of partners to more effectively combat foreign
actors attempting to infiltrate the social media space. Mr.
Gleicher noted that Facebook's engagement with their external
partners helped improve and refine the efficacy of their
detection and enforcement techniques. Mr. Gleicher concluded,
``By continuing to develop smarter technologies, enhance our
defenses, improve transparency, and build strong partnerships,
we are making the constant improvements we need to stay ahead
of our adversaries and to protect the integrity of our
platforms.''\174\
---------------------------------------------------------------------------
\174\Id.
---------------------------------------------------------------------------
Mr. Gleicher also described efforts and progress in
addressing inauthentic engagement on Instagram, which is owned
by Facebook. For example, Instagram penalizes accounts that are
associated with automated likes, comments, or follows to
artificially expand their reach. Using machine learning and
direct detection, the platform is able to ``identify accounts
that use third-party services to distribute inauthentic
engagement. When a service uses an account to generate
inauthentic activity, our tools can detect and remove that
activity before it reaches the recipient.''\175\ Instagram also
recently introduced the ability for community members to
directly report scams discovered on the platform. As with
Facebook, users are given more information about Instagram
accounts with substantial followings so that users can make
their own determination on the authenticity of the account.
This information includes the date when the account joined
Instagram, the country where the account is located, any
username changes in the last year, and any ads the account is
currently running.
---------------------------------------------------------------------------
\175\Id.
---------------------------------------------------------------------------
There have been several congressional inquiries into
Facebook's practices and policies in the aftermath of the 2016
election, and Facebook has undertaken certain new measures to
tighten its security and prevent similar abuse in 2020.
Facebook noted that it is testing new detection capabilities
that will help identify and remove accounts that impersonate
real people using their names or images. These new detection
processes can be particularly helpful in identifying fraudulent
accounts purporting to be some of the most frequently
impersonated members of the U.S. military and veterans'
community. The automated detection systems are trained to look
for certain techniques used by scammers to impersonate
individuals, such as omitting single letters of a person's name
to make the impostor account appear legitimate. Accounts that
are flagged for potential impersonation during the automated
review are then referred for additional human review. These
processes are intended to help more quickly detect impostor
accounts as soon as possible after creation and to remove them
immediately upon review and human verification, often before
people even see them.
Accounts and pages that claim false affiliation or
ownership with real organizations are unfortunately not limited
to veteran-related groups. ``In fact, the same bad actors
sometimes create multiple pages, some of which may impersonate
veterans'' organizations, while others might impersonate
organizations that focus on politically sensitive issues. That
is why, to root out and remove these bad actors, [Facebook]
focuses on patterns of behavior, not just content.''\176\
Facebook states that most removed accounts are blocked shortly
after creation, stemming the reach of the account before it can
do harm to other users or viewers. This approach allows
Facebook to be flexible to combat various types of
impersonation, and once Facebook develops effective tactics
with respect to one type of impersonation, they apply that
tactic to other types automatically.\177\
---------------------------------------------------------------------------
\176\Id. at 5.
\177\Id.
---------------------------------------------------------------------------
Facebook has told the Committee that it understands its
responsibility to ensure users, including veterans, are
protected from impersonation. Facebook also stated that it has
established dedicated escalation channels for individuals and
organizations most impacted by impersonation attempts,
including the Department of Defense. However, in response to
Rep. Jim Banks (R-IN) asking whether ``Facebook ha[d] a
specific process for reporting instances of veterans scamming
to federal law enforcement agencies,'' Mr. Gleicher did not
provide any specific procedures or resources applicable to
veterans.\178\
---------------------------------------------------------------------------
\178\HVAC Committee Hearing Transcript, at 76.
---------------------------------------------------------------------------
Is Facebook Doing enough?
Facebook has continued to draw attention and a measure of
criticism for its decisions to allow certain doctored content
on its platform that some users decry as deliberately
misleading or fake news. Compounding the problem, Facebook
partners with third-party fact-checkers to assess the veracity
of content and identify misinformation,\179\ but defers
substantially to the discretion of those external parties as to
what content is actually fact-checked. Thus, even content that
Facebook asserts is ``eligible'' to be fact-checked may not in
actuality be examined unless a third-party partner specifically
selects that content for review. The practical implication of
this structure is that Facebook is able to shift accountability
away from itself by pointing to its external fact-checking
partners, but then it does not appear to provide sufficient
guidelines on what content those partners must review--thereby
significantly eroding the efficacy of its fact checking
operations. Furthermore, Facebook has maintained its stated
policy that political speech and opinions from elected
officials, candidates or campaigns is not eligible for third-
party fact-checking.\180\ This seems to shift the burden of
verification from the platform onto users themselves. It is
questionable whether users have the awareness or means to
authenticate accounts or verify content on their own. Moreover,
because the social media platforms themselves have adopted
vastly disparate policies in terms of accepting political
advertisements, fact-checking, or identifying content that has
been challenged, users face an uneven and inconsistent social-
media landscape, where it becomes significantly harder to
determine reliability and trustworthiness.
---------------------------------------------------------------------------
\179\Facebook, Fact-Checking on Facebook, https://www.facebook.com/
business/help/2593586717571940?id=673052479947730.
\180\Facebook, Fact-Checking Program Policies, https://
www.facebook.com/business/help/315131736305613?id=673052479947730.
---------------------------------------------------------------------------
Facebook recently introduced an external, independent
review appeals board that functions in a quasi-judicial
capacity to review certain of Facebook's content decisions and
allow users to seek an additional layer of review for
challenges to removed content.\181\ However, it does not appear
that this review board will have any access to authentication
data for accounts or posts to help determine the legitimacy of
users or content--but rather will function solely as the final
arbiter of certain content moderation decisions. This is of
decidedly less importance to the issue of spoofing and the
distribution of disinformation or propaganda than it would be
to potential claims of content standard violations such as
decency/pornography claims, etc.
---------------------------------------------------------------------------
\181\Josh Constine, Facebook's New Policy Supreme Court Could
Override Zuckerberg, TechCrunch (Sept. 17, 2019), https://
techcrunch.com/2019/09/17/facebook-oversight-board/?guccounter=1.
---------------------------------------------------------------------------
Moreover, there is no appeals process that would enable a
user such as VVA to elevate claims of a misappropriated or
fraudulent account to a higher body for expedited review.
During the October hearing, Rep. Gus Bilirakis (R-FL) noted the
trouble that Mr. Goldsmith encountered in reporting and trying
to take down the spoofed site.\182\ Mr. Goldsmith stated that
he repeatedly and persistently sought to bring the spoofed VVA
account to Facebook's attention and still faced undue delays, a
lack of transparency, and a lack of direct communication to
help get the fraudulent account shut down expeditiously.\183\
---------------------------------------------------------------------------
\182\HVAC Committee Hearing Transcript, at 54-55.
\183\Mr. Goldsmith written testimony, at 6.
---------------------------------------------------------------------------
During the Committee hearing, Rep. Conor Lamb (D-PA) asked,
``How much does Facebook spend on this specific problem set, in
terms of paid employees, investments in the AI, and tech
tools?''\184\ The response from Facebook was that on ``the
overall problem . . . [Facebook] ha[s] more than 35,000
employees working in this space. [Facebook] currently spend[s]
more money today each year than the company made in profits the
year that it IPO'd. Very, very large amounts.''\185\
---------------------------------------------------------------------------
\184\HVAC Committee Hearing Transcript, at 60.
\185\Id.
---------------------------------------------------------------------------
When Rep. Joe Cunningham (D-SC) posed a related question,
Facebook stated that there are 35,000 employees working on
safety and security generally and this number is triple what it
was a few years ago.\186\ Rep. Lauren Underwood (D-IL) queried
both Facebook and Twitter about the general timeline for
someone who lodges a complaint to be able to communicate with
an actual person, but neither company provided a definitive
answer, noting instead that it depended on the specific
circumstances and the manner of report.\187\
---------------------------------------------------------------------------
\186\Id. at 48-49.
\187\Id. at 69.
---------------------------------------------------------------------------
The two primary areas in which Facebook has opportunities
to do more to combat spoofing are verification of its own
accounts and content and sharing more robust data with its peer
platforms and law enforcement agencies.
More stringent review or verification of authentication
data for new accounts would directly help reduce spoofing by
making it harder to create fraudulent or misleading pages.
Facebook already imposes higher verification standards for
pages and groups with large audiences.\188\ Expanding that
level of review for all accounts, and including geolocation
information for owners, should be a feasible step.\189\
Although Facebook has outlined its efforts, investments, and
initiatives designed to review and remove inauthentic content,
it has notably excluded the significant category of political
speech and opinion, including within paid advertisements, by
candidates and campaigns from such processes.\190\ Given the
significant use of spoofing to seek to influence elections,
political advertisements and communications are a prime
opportunity for Facebook to adopt stronger enforcement
practices. Facebook has made an incremental step in this
direction by issuing a refined policy on political
disinformation regarding the 2020 election.\191\
---------------------------------------------------------------------------
\188\Sarah Perez, Facebook and Instagram Will Now Show Location of
Posts from High-Reach Accounts Targeting US Audiences, TechCrunch (Apr.
22, 2020), https://techcrunch.com/2020/04/22/facebook-and-instagram-
will-now-show-location-of-posts-from-high-reach-accounts-targeting-u-s-
audiences/.
\189\Id.
\190\Facebook, Fact-Checking Program Policies, https://
www.facebook.com/business/help/315131736305613?id=673052479947730; Tony
Romm & Isaac Stanley-Becker, Tens of Thousands of Political Ads on
Facebook Lacked Key Details About Who Paid for Them, New Report Finds,
The Washington Post (Mar. 8, 2020), https://www.washingtonpost.com/
technology/2020/03/06/tens-thousands-political-ads-facebook-lacked-key-
details-about-who-paid-them-new-report-finds/.
\191\Jessica Guyun, Facebook Bans Ads that Seek to Delegitimize the
Election or Make False Claims About Voting, USA Today (Sept. 30, 2020),
https://www.usatoday.com/story/tech/2020/09/30/facebook-bans-ads-
delegitimize-election-make-false-voting-claims/5875095002/.
---------------------------------------------------------------------------
Similarly, sharing more comprehensive data about ownership,
authentication, and activity patterns in instances of spoofing,
fraud, or criminal activity would enable law enforcement and
peer platforms to respond more efficiently and comprehensively
in identifying bad actors.\192\ Such measures would need to be
carefully crafted to protect user privacy and civil rights
concerns. Creating a law enforcement exclusion in the platform
terms of service agreements for users could be a potential
solution that balances privacy interests with law enforcement
needs. Additionally, to the extent permitted within the
existing legal structure, Facebook should increase the
frequency of notification and the scope of data exchanged with
its peer platforms. Specifically, information related to
fraudulent and removed accounts should be exchanged in order to
facilitate identification and removal of related accounts on
those other platforms. Facebook should also disclose more
information about the frequency and nature of its
communications with its peer platforms and law enforcement,
including the scope and detail of the data that is shared about
identified foreign infiltrators and spoofers.\193\ The efforts
undertaken to increase transparency and responsiveness to
complaints simply are not enough to prevent this threat from
spreading, nor do the changes address the issues that VVA
experienced in requesting removal of fraudulent pages.
---------------------------------------------------------------------------
\192\Paul M. Barrett, NYU Stern Center for Business and Human
Rights, Disinformation and the 2020 Election: How the Social Media
Industry Should Prepare (2019), https://issuu.com/
nyusterncenterforbusinessandhumanri/docs/nyu_election_2020_report?
fr=sY2QzYzI0MjMwMA; see also Queenie Wong, Facebook's Transparency
Efforts Around Political Ads Fall Short, Study Finds, CNET (Mar. 6,
2020), https://www.cnet.com/news/facebooks-transparency-efforts-around-
political-ads-fall-short-study-finds/.
\193\Id.
---------------------------------------------------------------------------
TWITTER
Twitter is a major American social media, networking, and
microblogging service, with 186 million daily active users as
of June 30, 2020.\194\ In his appearance before the Committee,
and in response to a question from Rep. Andy Barr (R-KY),
Twitter's Public Policy Manager, Kevin Kane, said, ``Every day
there are more than 500 million tweets around the world on
Twitter. And as I mentioned, we actioned approximately 335,000
accounts that were permanently suspended that were engaging in
scamming activity.''\195\ On Twitter, users post and interact
with brief written messages known as ``tweets.'' Tweets can
also incorporate images, videos, links to articles, or other
media into the messages. By redistributing (retweeting)
messages broadly across subsequent networks, Twitter users
amplify the messaging of the original tweet. Registered users
can post, ``like,'' and retweet tweets, but unregistered users
can only read them.
---------------------------------------------------------------------------
\194\Twitter, Q2 2020 Letter to Shareholders at 2 (July 23, 2020),
https://s22.q4cdn.com/826641620/files/doc_financials/2020/q2/Q2-2020-
Shareholder-Letter.pdf.
\195\HVAC Committee Hearing Transcript, at 92.
---------------------------------------------------------------------------
How Twitter is Combatting Spoofing
In the aftermath of the 2016 U.S. election, Twitter
received significant scrutiny for the role it may have played
in shaping and driving American political discourse and
opinion, and particularly with respect to the then-emerging
idea of fake news and misinformation.\196\
---------------------------------------------------------------------------
\196\Daisuke Wakabayashi & Scott Shane, Twitter, With Accounts
Linked to Russia, to Face Congress Over Role in Election, The New York
Times (Sept. 27 2017), https://www.nytimes.com/2017/09/27/technology/
twitter-russia-election.html.
---------------------------------------------------------------------------
In the fall of 2017, Twitter undertook an analysis of how
its platform, networks, and technology may have been
deliberately manipulated by foreign actors for the purpose of
influencing the election through the dissemination of political
propaganda or socially divisive content. The analysis included
both an investigation into activity specifically by the Russian
Internet Research Agency, and a broader inquiry into all
malicious automated activity (posting, ``liking,'' or
retweeting) originating in Russia. Twitter also reviewed a
comprehensive collection of election-related Tweets from
accounts linked to Russia, and compared the activity levels of
those selected accounts to overall activity levels on
Twitter.\197\ Mr. Kane testified that this analysis found
50,258 automated accounts that were Russian-linked and tweeting
election-related content, representing less than two one-
hundredths of a percent (0.016%) of the total accounts on
Twitter at the time.\198\ These accounts generated 2.12 million
tweets, or approximately one percent of the total volume of
election-related Tweets, during that period. Twitter also
analyzed accounts that paid for advertisements promoting
election-related Tweets over the course of 2016 and discovered
only nine such accounts with ties to Russia.\199\
---------------------------------------------------------------------------
\197\Hijacking Our Heroes: Exploiting Veterans through
Disinformation on Social Media Before the H. Comm. On Veterans'
Affairs, 116th Cong. (2019) (written testimony of Mr. Kevin Kane,
Public Pol'y Mgr., Twitter, at 3-4) (access Mr. Kane's written
testimony here, https://docs.house.gov/meetings/VR/VR00/20191113/
110183/HHRG-116-VR00-Wstate-KaneK-20191113.pdf).
\198\Id.
\199\Id.
---------------------------------------------------------------------------
Upon identifying and isolating the account data associated
with Russia's Internet Research Agency (IRA), Twitter published
a data set of removed accounts and underlying data (e.g.,
message contents) that were from state-backed foreign entities
(including the IRA associated data). This data set has been
studied by law enforcement, peer platforms, and outside
analysts, including Graphika. Twitter released the full,
comprehensive archives of Tweets and media associated with
potential information operations found on the platform,
including 3,613 accounts believed to be associated with the IRA
dating back to 2009. Twitter encouraged open research and
investigation of these datasets by researchers and academics in
order to identify potential behavioral patterns that might help
improve deterrence protocols. Prior to the release of these
datasets, Twitter shared individual examples of alleged foreign
interference by the IRA in political conversations on the
platform. Twitter also provided direct notice to any users if
they had interacted with any of these state-backed accounts. As
stated by Mr. Kane, ``[Twitter] launched this unique initiative
to improve academic and public understanding of these
coordinated campaigns around the world, and to empower
independent, third-party scrutiny of these tactics on our
platform.''\200\
---------------------------------------------------------------------------
\200\Id.
---------------------------------------------------------------------------
Twitter continues to maintain a public archive of removed
accounts. It claims that this archive is now the largest of its
kind in the industry, and that thousands of researchers have
used these datasets that contain more than 30 million
individual Tweets and more than one terabyte of media.\201\
Twitter also periodically publishes new datasets of removed
accounts (but not the underlying content) and without any
associated signals data that would enable other platforms, law
enforcement, or analysts to trace activity from the same
foreign entities across platforms, or to other accounts on the
same platform. Instead the currently available data represents
isolated static snapshots of fraudulent accounts that have
already been removed by the time the data is made available.
---------------------------------------------------------------------------
\201\Id. at 5.
---------------------------------------------------------------------------
Unfortunately, data included in the public archive is of
very limited practical use for law enforcement, analysts, or
think tanks in terms of trying to predict future activity
patterns or understand foreign network breadth for prospective
deterrence, and is similarly limited for other platforms
seeking to identify and deter known actors before they are able
to infiltrate their own platforms.\202\ Once an account has
been removed, important signal data like internet protocol (IP)
address, geolocation, or timing of account activity can no
longer be used to actively trace where a user is operating
from, what other accounts use the same IP address, or whether
accounts on other platforms share any of the same signals data
(which might indicate that a given user holds accounts on
multiple platforms).
---------------------------------------------------------------------------
\202\HVAC Interview with Dr. Vlad Barash on September 25, 2019.
---------------------------------------------------------------------------
Twitter maintains that it is restricted from sharing the
underlying content, even from accounts that have been removed
on the theory that the privacy protections under its Terms of
Service and Privacy Policy are extended even to fraudulent or
removed accounts.\203\ As part of the Terms of Service which
govern Twitter's relationship with its users, Twitter includes
a written Privacy Policy.\204\ The Privacy Policy lays out the
terms and scenarios under which Twitter shares private user
data with any outside parties, including third-party service
providers, advertisers, and law enforcement. Private data
includes any information that the user does not share publicly
(e.g. direct messages or protected tweets), is not required to
be shared for basic operation of Twitter (e.g. with service
providers or advertisers), or is not otherwise authorized by
the user. Twitter allows users to control when most of their
own private data can be shared, but identifies certain
exceptions including, notably, ``law, harm, and the public
interest.'' Specifically, Twitter states that it may ``disclose
your personal data or other safety data if we believe that it
is reasonably necessary to comply with a law, regulation, legal
process, or governmental request.''\205\
---------------------------------------------------------------------------
\203\HVAC Interview with Kevin Kane on October 17, 2019.
\204\Twitter, Privacy Policy, https://twitter.com/en/privacy.
\205\Id.
---------------------------------------------------------------------------
In response to a request by Committee staff for data
related to removed accounts, suspected bot accounts, and direct
messages, Twitter explained that it requires legal process to
turn over such information. Twitter further stated that under
Title II of the Electronic Communications Privacy Act (ECPA),
also known as the Stored Communications Act,\206\ legal process
is required for the disclosure of all private data, and that
any voluntary disclosure by Twitter without a specific legal
requirement would violate its Privacy Policy.\207\ In the
absence of any legal authority or statutory exemption
specifically compelling the production of private data without
a subpoena (even for law enforcement or regulatory purposes),
disclosure from Twitter, other social media platforms, and
internet service providers generally requires a time-consuming
legal process that hampers the ability of law enforcement to
use such data in an expedited manner to identify, obstruct, or
apprehend the offenders. Although Twitter concedes that
spoofing and misrepresentation violate its Terms of Service, it
believes that its legal obligation to the user under the
Privacy Policy continues in force even though the account may
be removed for those violations. The specific legal question of
whether Title II of ECPA continues to protect data in cases of
fraud or illegal activity is not clear based on legal
precedent.
---------------------------------------------------------------------------
\206\18 U.S.C. Sec. 2701-2713.
\207\18 U.S.C. Sec. 2702; HVAC Interview with Kevin Kane and Stacia
Cardille on September 10, 2019.
---------------------------------------------------------------------------
On October 30, 2019, Twitter announced a new global policy
to stop all political advertising.\208\ Twitter defined
political advertising to include any paid messaging that
references a candidate, political party, government official,
ballot measure, or legislative or judicial outcome.\209\ The
policy is based on the belief that ``earned sharing'' of
political messaging (as measured by retweets and likes) is
better and more organic than purchasing political advertising.
Twitter Chief Executive Officer Jack Dorsey has stated that
``paying for reach removes that decision, forcing highly
optimized and targeted political messages on people.''\210\
Dorsey reasoned that targeted ads ``present entirely new
challenges to civic discourse: machine learning-based
optimization of messaging and micro-targeting, unchecked
misleading information, and deep fakes. All at increasing
velocity, sophistication, and overwhelming scale.''\211\
Candidates, campaigns and parties are still able to share
content, but they cannot simply extend the reach of that
content through paid advertising. Twitter's decision to ban
paid political advertisements has been roundly commended.
---------------------------------------------------------------------------
\208\Kate Conger, Twitter Will Ban All Political Ads, C.E.O. Jack
Dorsey Says, The New York Times (Oct. 30, 2019), https://
www.nytimes.com/2019/10/30/technology/twitter-political-ads-ban.html.
\209\Twitter, Political Content Policy, https://
business.twitter.com/en/help/ads-policies/prohibited-content-policies/
political-content.html; see also Kate Conger, What Ads are Political?
Twitter Struggles With a Definition, The New York Times (Nov. 15,
2019), https://www.nytimes.com/2019/11/15/technology/twitter-political-
ad-policy.html.
\210\Kate Conger, Twitter Will Ban All Political Ads, C.E.O. Jack
Dorsey Says, The New York Times (Oct. 30, 2019), https://
www.nytimes.com/2019/10/30/technology/twitter-political-ads-ban.html.
\211\Id.
---------------------------------------------------------------------------
However, there are still loopholes which facilitate the
promotion of political agendas without conflicting with the ad
ban. Messages can be crafted around political issues without
naming specific candidates, parties, or outcomes.\212\
Additionally, while Twitter no longer allows for ads to be
targeted as narrowly as by ZIP code, targeting based on a
user's state or province is still possible.\213\
---------------------------------------------------------------------------
\212\Id.
\213\Barbara Orutay, Twitter Details Political Ad Ban, Admits It's
Imperfect, AP News (Nov. 15, 2019), https://apnews.com/
5dc8154a740649c89ec3c4e9bc5fba0f.
---------------------------------------------------------------------------
Mr. Kane further testified that Twitter has specific
guidelines that govern a user's ability to share information
about elections. He noted that users are prohibited from
posting false or misleading information about how to
participate in an election, including information about how to
vote or voter registration, voter identification requirements,
and the date or time of an election. Additionally, users may
not attempt to intimidate or dissuade voters from participating
in an election by sharing false or misleading information,
including claims about polls being closed, long lines, voting
equipment issues, votes not being counted, or law enforcement
activity around poll sites. Finally, Mr. Kane also noted that
Twitter does not allow ``the creation of fake accounts which
misrepresent their affiliation or share content that falsely
represents its affiliation to a candidate, elected official,
political party, electoral authority, or government
entity.''\214\
---------------------------------------------------------------------------
\214\Mr. Kane written testimony, at 6.
---------------------------------------------------------------------------
Is Twitter Doing Enough?
Twitter provided the Committee with only broad details on
how the platform reviews content to screen for potential
violations or coordinated activity. Twitter told Committee
staff that it uses a layered review process similar to the one
used by Facebook, incorporating both artificial intelligence/
automated machine review and human assessment. Content reported
by users for potential violations of platforms is all reviewed
by human content moderators, and there is a well-defined
appeals process.
In an exchange with Rep. Gilbert Cisneros (D-CA) during the
hearing, Mr. Kane stated, ``[Twitter] continues to invest and
look at the behavior, look at the signals behind how these
accounts are behaving and potentially targeting people, to
include veterans. But again, we take a much more holistic
approach so we are not just silencing certain communities, and
we can apply lessons learned across the board. But again, it is
looking at the signals behind the accounts, as well as
potential coordinated behavior, which is a very strong signal
that accounts are engaging in suspicious activity and cause us
to look into it further.''\215\ In response to a direct
question from Mr. Cunningham, Twitter testified that it has
devoted 4,700 persons to content moderation.\216\ Ms. Underwood
inquired specifically about the ability of a victim to engage
content reviewers by telephone, but Mr. Kane noted that users
are not able to do so presently.\217\
---------------------------------------------------------------------------
\215\HVAC Committee Hearing Transcript, at 89.
\216\Id. at 48.
\217\Id. at 60.
---------------------------------------------------------------------------
Twitter's testimony about its internal investigative
approach and how these complex, sometimes cross-jurisdictional
operations are identified was presented in broad generalities
that obscured the particulars of the type and scope of
information that is shared with peer platforms and law
enforcement (although it was repeatedly mentioned that such
cooperation and collaboration does occur). Twitter recognizes
that, as a private company, there are threats that it cannot
understand and address alone. Twitter has disclosed that it
participates in regular discussions with law enforcement and
other platforms, including quarterly briefings with FITF on
foreign influence. Twitter also meets monthly with
representatives from FBI, DHS, ODNI, DOJ, and industry peers to
discuss 2020 election threats. Starting in 2018, a number of
technology companies--including Twitter--established a
dedicated, formal communications channel to facilitate real-
time information sharing regarding election integrity, and
Twitter continues to utilize that communications channel for
ongoing information sharing. Twitter did not provide any
details on the nature or scope of data exchanged, or other
systemic details.\218\ Nor did Twitter's testimony describe the
nature of the communications between Twitter and its peers or
law enforcement, and the scope and detail in the data that is
shared about identified foreign infiltrators and spoofers.\219\
The lack of data sharing represents a significant impediment to
determining when foreign-based actors might be launching
infiltration attacks across multiple platforms, or to
anticipate such attacks in a timely manner to effectively
minimize potential harm.
---------------------------------------------------------------------------
\218\Id. at 47-48, 102.
\219\Id. at 47-48.
---------------------------------------------------------------------------
Twitter continues to monitor and enforce political accounts
for compliance with its Rules on Platform Manipulation. This
was recently enforced against the campaign of Michael Bloomberg
in the Democratic primaries, resulting in the suspension of
seventy accounts for coordinated behavior.\220\ However,
opportunities remain for spoofers to exploit gaps between these
policies, for example, by using divisive content that does not
meet the threshold of paid political advertising, but serves
similar purposes. For example, Twitter still allows ads related
to social causes such as climate change, gun control, and
abortion, but organizations cannot advocate for or against a
specific political, judicial, legislative, or regulatory
outcome related to those matters.\221\
---------------------------------------------------------------------------
\220\Meryl Kornfield, Twitter Suspends 70 Pro-Bloomberg Accounts
for Campaign's Copy-and-Paste Strategy, The Washington Post (Feb. 22,
2020), https://www.washingtonpost.com/politics/2020/02/22/bloomberg-
twitter-suspensions/.
\221\Emily Stewart, Twitter Is Walking into a Minefield with Its
Political Ads Ban, Vox (Nov. 15, 2019), https://www.vox.com/recode/
2019/11/15/20966908/twitter-political-ad-ban-policies-issue-ads-jack-
dorsey.
---------------------------------------------------------------------------
Given that there are strict restrictions on the ability of
foreign actors to contribute to or coordinate with U.S.
political campaigns, either the platforms must monitor
themselves or empower federal regulators to police the
platforms for potential violations of U.S. election laws. Both
Twitter and Facebook acknowledged in their comments to Mr.
Cunningham that they report their findings and takedowns to the
Federal Election Commission and the Federal Bureau of
Investigation, when the offending patterns of activity might
indicate foreign efforts to influence political messaging or
elections. Twitter published a report of its findings from the
2018 U.S. midterm elections. The 2018 U.S. midterm elections
were the most Tweeted-about midterm election in history with
more than 99 million Tweets sent from the first primaries in
March through Election Day.\222\
---------------------------------------------------------------------------
\222\Mr. Kane written testimony, at 4.
---------------------------------------------------------------------------
Twitter previously offered a ``verified user'' feature
which allowed certified authentic accounts to be identified
with a blue checkmark next to their posts. The verification
feature has now been suspended for new users with a few narrow
categories of exceptions, such as celebrities, journalists, and
public officials. Twitter has said that it is committed to
verifying VSOs, but the reality is that it is still difficult
for these organizations to receive verification. Even a
Congressionally chartered VSO such as VVA experienced
difficulties and delay in trying to get a verified account. As
Mr. Goldsmith noted during his testimony, the only reason that
VVA was able to ultimately obtain a verified account on Twitter
was because he facilitated the request through a personal
relationship with a Twitter employee.\223\ Twitter has informed
HVAC that all VSOs with Twitter accounts have now been
verified, and has committed to working with the committee to
ensure that congressionally-chartered VSOs, and their
affiliated chapters, continue to be verified.
---------------------------------------------------------------------------
\223\HVAC Committee Hearing Transcript, at 55.
---------------------------------------------------------------------------
As foreign spoofers become more sophisticated in their
ability to impersonate or imitate legitimate accounts and
users, each individual piece of available signals data becomes
more valuable in the effort to identify digital fingerprints in
order to efficiently intercept (if not prevent) such attacks.
While it is unclear what data it may provide through private
channels to other platforms or to regulators, Twitter's
compliance with the procedures required under Title II of ECPA
before releasing data to the broader public presents an
impediment to combatting foreign spoofing in an efficient and
timely manner. Ranking Member Dr. David P. Roe stated, ``We
must empower veterans with the information necessary to make an
informed choice about whether the benefits of social media are
worth the risks and to make them aware of available resources
to protect themselves.''\224\
---------------------------------------------------------------------------
\224\Id. at 10.
---------------------------------------------------------------------------
Twitter should restore its verification process for
accounts so that users are able to quickly and easily discern
which accounts have been reviewed and vetted. Additionally,
Twitter should disclose information about the nature and
frequency of its communications and data sharing with its peer
platforms and with law enforcement. Importantly, Twitter should
support a revised process by which data tied to inauthentic or
criminal behavior can be efficiently and adequately shared with
other platforms and law enforcement so that spoofers are not
able to jump from platform to platform to elude enforcement.
However, all such data sharing measures must be designed to
balance and protect users' privacy interests as well.
The Role of Law Enforcement--Briefing with Committee
On January 14, 2020, representatives from the FBI briefed
Committee Members and staff on the growing issue of spoofing
targeting veterans. The meeting was bipartisan and on-the-
record.\225\ FBI participants were drawn from the Foreign
Influence Task Force (FITF) and the Criminal Investigative
Division (CID), including CID staff of the Money Laundering,
Forfeiture, and Bank Fraud Section, the Financial Crimes
Section, and the Economic Crimes Unit. Section Chief Brad
Benavides of FITF and Acting Deputy Assistant Director Steve
Merrill of CID, the two senior members of the panel, provided
the majority of the comments on behalf of the FBI. CID handles
the FBI's efforts to identify, deter, and disrupt significant
complex financial, health care, money laundering, and
intellectual property crime threats impacting the United
States. The FITF is a multi-divisional/multi-agency task force
comprised of agents, analysts, task force officers, and
professional support focused on combating malign foreign
influence efforts targeting the United States. The CID is ``the
largest major division in the Bureau, supporting more than
4,800 Special Agent[s]. Within this division, the emphasis is
on preventing crimes related to national security, such as
interrupting financial networks that provide support to
terrorist organizations. A large number of personnel are also
allocated to violent crimes, financial crimes, and organized
crime.''\226\
---------------------------------------------------------------------------
\225\Remarks and statements from law enforcement are from HVAC
Round Table Discussion on January 14, 2020.
\226\Careers for FBI Special Agents in the Criminal Investigations
Division, FBIAgentEDU.org, https://www.fbiagentedu.org/careers/fbi-
special-agent/special-agent-
criminal-investigations/ (last visited Sept. 16, 2020).
---------------------------------------------------------------------------
Both FITF Section Chief Benavides and Acting Deputy
Assistant Director of CID Merrill used their opening remarks to
highlight the target rich environment that an aging veteran
demographic provides for potential criminals. CID emphasized
that veterans are attractive targets for financial exploitation
and scams due to their steady income streams from pensions,
annuities, and VA benefits payments. The other sections chiefs
also acknowledged the serious threats that veterans face from a
multitude of spoofing attacks, ranging from romance scams to
commercial fraud to replicated websites being used to advance
misinformation campaigns.
THREAT EVALUATION AND STATISTICS
The overall volume of internet crime complaints is a
staggering $3.5 billion in aggregate annual losses, which
includes the categories of commercial fraud, international
campaigns, and romance scams. FBI representatives described a
massive increase in romance scams, which notably grew from $211
million in 2017\227\ to $362 million in 2018\228\ and to $475
million in 2019.\229\ The scale of the problem is likely even
larger than those numbers might suggest because the FBI noted
that these scams are often underreported due to victims'
embarrassment or reluctance to disclose to their families that
they have been scammed or lost money.\230\
---------------------------------------------------------------------------
\227\Federal Bureau of Investigation, U.S. Dep't of Justice, 2017
Internet Crimes Report at 21 (on file at https://pdf.ic3.gov/
2017_IC3Report.pdf).
\228\Federal Bureau of Investigation, U.S. Dep't of Justice, 2018
Internet Crimes Report at 20 (on file at https://pdf.ic3.gov/
2018_IC3Report.pdf).
\229\Federal Bureau of Investigation, U.S. Dep't of Justice, 2019
Internet Crimes Report at 20 (on file at https://pdf.ic3.gov/
2019_IC3Report.pdf).
\230\Remarks and statements from law enforcement are from HVAC
Round Table Discussion on January 14, 2020.
---------------------------------------------------------------------------
Putting the threat to veterans in perspective, Ranking
Member Roe asked the law enforcement representatives about the
effectiveness of these spoofing schemes. The FBI stated that
spoofing is effective because, put simply, it works and that is
precisely why malign actors use it as a technique. The FBI
representatives mentioned that one of the surest ways to limit
the reach of spoofers is to improve cyber-hygiene, or the
practices and precautions that users of computers and social
media should take to maintain system health and improve online
security. These practices are often part of a routine to ensure
the security of identity, maintenance of privacy, and
protection of personal data including financial or identity
details.\231\ Examples of such practices include keeping
passwords secret, not divulging banking or credit card
information carelessly, and being vigilant for attempted
impersonation.
---------------------------------------------------------------------------
\231\Nadia Kovacs, Tips for Protecting Your Social Media Privacy,
NortonLifeLock, https://us.norton.com/internetsecurity-privacy-
protecting-privacy-social-media.html (last visited Sept. 16, 2020).
---------------------------------------------------------------------------
On the issue of investigating foreign actors seeking to
distribute political propaganda or influence American
elections, the FBI relies upon the Foreign Agent Registration
Act (FARA) for authority and access to investigative tools to
pursue these perpetrators and networks overseas. The specific
tools, however, are not available in matters that do not
involve foreign actors, and therefore the FBI must rely on a
range of less effective strategies to investigate and eliminate
other forms of spoofing, including the voluntary sharing of
data by the social media platforms and anecdotal reports of
potential fraud. Although CID and FITF have responsibilities
for different aspects of spoofing, and consequently have access
to different tools, both agencies agreed on the importance of
receiving data and information regarding potential crimes in a
timely manner.
Both CID and FITF also noted that one of the most effective
tools to eliminate inauthentic online behavior including
spoofing, is the Terms of Service (TOS) implemented by the
respective platforms for their users. Violations of those terms
by bad actors, through impersonation, spam, fraud, intellectual
property violations, or other prohibited conduct, enables the
platforms to suspend or terminate the offending accounts, and
to remove content from the platform archives in certain cases.
Enforcement of the TOS by the platforms is the most efficient
and expeditious way to remove violators and their content from
a given platform because the platform is the ultimate arbiter
of its own rules, and such enforcement does not require the
participation of law enforcement.
However, when a violation of TOS may also involve criminal
activity, such as fraud or actions by a foreign actor, the FBI
strongly emphasized the need for immediate communication
between the platforms and law enforcement. The FBI
representatives specifically noted the importance of quickly
sharing the details and underlying identification data about
the accounts undertaking the spoofing or fraud when the
platforms take down such accounts. The FITF representatives
further suggested that when social media companies identify
inauthentic behavior on their platforms, they should
immediately notify and engage law enforcement before taking
down the offending accounts. Mr. Benavides stated that this
would allow the FBI to monitor, trace, observe and gather
relevant data from the active accounts, which will directly
assist in tracing foreign networks, individual accounts across
platforms, or specific bad actors running multiple accounts.
None of these outcomes would be feasible if law enforcement is
only apprised after the fact (i.e., after the platforms have
already taken down the bad accounts). In the absence of sharing
such contemporaneous notice and underlying data, the FBI noted
severe limitations on its ability to retroactively investigate
or identify bad actors, let alone develop more robust systems
or tools for intercepting such campaigns in the future.
COMMUNICATIONS AND DATA SHARING
When Chairman Mark Takano asked about the current nature
and scope of the engagement with the platforms, the two FBI
components described starkly different relationships with the
platforms. Section Chief Benavides described what he considers
to be a strong working relationship and general satisfaction
with the social media firms. FITF holds standing quarterly
meetings with social media companies, which cover a broad range
of topics covering inauthentic behavior on the platforms,
including spoofing, spamming, and bot activity. He described
the relationship as positive because the voluntary free flow of
information that FITF enjoys allows his section to assess and
work through threats with social media companies. Section Chief
Benavides felt that the voluntary information exchange provides
for a better working relationship as opposed to social media
companies providing information on an ad hoc request-by-request
basis. Mr. Benavides reasoned that the informal working
relationships that develop between FITF personnel and the
enforcement teams at the social media platforms facilitates a
more robust discussion of issues that would perhaps otherwise
not rise to the threshold of a formal reporting requirement or
an actual legal violation. This unstructured discussion of
potentially suspect activity, instead of the higher bar of
actual crimes or specific conduct, enables FITF to engage the
social media platforms earlier than might be feasible in a more
formal reporting environment.
On the other hand, the CID's Financial Crimes team noted
much less satisfaction with its current relationships with the
social media platforms and drew a marked contrast between those
relationships and the one that CID enjoys with the highly
regulated banking sector. CID described numerous, regular
contacts held between the FBI (in conjunction with the
Department of the Treasury) and the banking sector. The
relationship and communications between the FBI and the banking
sector include quarterly national meetings with the big banks
and financial institutions. Furthermore, the FBI headquarters
in Washington D.C. encourages each of the 56 field offices to
engage directly with financial institutions in their respective
territories. Unlike FITF, CID does not have any such regularly
scheduled meetings with the social media platforms.
CID often receives information about social media incidents
informally through individual relationships between people at
social media companies and FBI personnel, which is generally
less helpful than the formal, systematic reporting structure in
place with the financial sector. The most notable source of
reporting about online fraud is from the victims who have lost
money themselves and report directly to CID, rather than from
the social media companies. The piecemeal anecdotal information
creates a lack of uniformity in timing, detail, and data. This,
in turn, hampers the ability of CID to form a systemic
understanding of the problems or solutions, and from
synthesizing aggregate data that could potentially be used to
identify patterns, trace networks, or otherwise develop
comprehensive defensive protocols. Particularly with respect to
fraud or financial crimes, reliance on anecdotal evidence also
prevents CID from understanding the full scale of potential
crimes because they cannot accurately compile aggregated data
without detailed information from the platforms themselves.
Relying on individual victim reports cannot substitute for
the more comprehensive information that would be available from
the platforms. As a result, CID described its own actions as
generally reactive, instead of proactive, due to this lack of
detailed, aggregated information and formal reporting of
suspected criminal activity. If a platform takes down certain
accounts, CID is only made aware of such action if the platform
voluntarily notifies it--there is no alternative way for CID to
track that information independently. Broad categories of
platform takedowns should arguably trigger automatic reporting,
for example, when a platform acts in response to criminal
activity such as fraud or child-welfare issues, but even those
disclosures are currently voluntary. Additionally, CID noted
that it cannot know what additional details the social media
companies might be withholding from disclosure. For example,
platforms may be reticent to divulge additional details that
might expose platform vulnerabilities until after the platform
is able to address such vulnerabilities.
IS LAW ENFORCEMENT DOING ENOUGH?
Throughout the discussion the panel discussed the lack of
legislative or statutory disclosure requirements. Law
enforcement relies almost entirely on social media platforms'
voluntary disclosures and cooperation to take down actors
conducting spoofing operations. While the two components, FITF
and CID, were split in their current satisfaction with the
level of cooperation and communication received from the social
media platforms, both were also hesitant to recommend potential
changes to the statutory framework. Representatives from both
FITF and CID expressed some reticence toward creating a
statutory obligation for reporting or standards because such an
endorsement might jeopardize the current relationships and
voluntary information exchanges, or inadvertently create a
higher triggering threshold before formal reporting is required
than is currently enjoyed--specifically by the FITF. However,
both components agreed on the importance of timely
communications and access to comprehensive data. So, while a
mandatory reporting structure may not be advisable, any
measures aimed at facilitating law enforcement's access to data
in an expedited manner would greatly enhance their ability to
identify and isolate potential criminals.
CID also recommended aggregation of the data being reported
and explained that in the banking industry there are
``suspicious activities reports'' prepared for and provided to
the Department of the Treasury, which then shares that
information with the Financial Crimes Section. That mechanism
can be used as a model to aggregate and report data on fraud in
the social media arena. There is a mandatory centralized
repository of data for banking reports, and it would be very
beneficial if CID had access to a comparable system containing
reports from social media platforms. This would be a
significant departure from the current practice in which CID
only sporadically learns about fraudulent activities through
anecdotal victim reports, which are notably underreported,
supplemented by irregular voluntary communications from the
platforms themselves. In response to a question about potential
mandatory reporting requirements for social media companies,
CID representatives indicated that when the FBI prosecutes
someone successfully, it announces it because it is important
for the public to be knowledgeable about enforcement actions on
the social media platforms and the consequences of perpetrating
fraud. However, due to the mutually beneficial relationship
experienced by the FBI and social media platforms, the
representatives generally expressed desire to continue the
current voluntary information exchange rather than potentially
jeopardize the existing streams of information.
The CID representatives noted that their work with the
banking sector could be used as a model for creating a better
formal working relationship with the social media companies. A
similar combination of regular meetings, along with an agreed
upon protocol for reporting the discovery of fraud or criminal
activity and information related to removed accounts, would be
an advisable addition to the current working relationship
between CID and the social media platforms. CID emphasized that
these changes to the current operating procedures regarding
notification, reporting, data exchange, and communication
between the social media companies and the FBI would increase
the division's effectiveness, ability to prosecute, and
deterrence of spoofing incidents. Interestingly, given the
significant disparities in how the social media platforms
communicate and exchange information with CID as compared to
FITF, it appears that much of this problem is simply based on
the internal policies of the social media platforms, rather
than rooted in any systemic obstacle.
When considering potential improvements to address spoofing
issues and the threats they pose to veterans, the FBI panel had
five recommendations:
(1) create a standard of aggregation for both
reported violations of the terms of service (TOS) and
unreported (but acted upon) violations of TOS;
(2) establish a universal, industry-wide standard for
releasing detailed data from account takedowns, and
facilitating access for law enforcement to obtain such
data in a timely manner;
(3) improve the communication by social media
platforms of suspicious activities to federal law
enforcement prior to takedown in order to enable social
media platforms and law enforcement to work in tandem
to address spoofing threats and identify foreign
networks;
(4) establish a standard for reporting suspected
spoofed accounts/pages to law enforcement prior to
removing or taking down the accounts/pages; and
(5) start an education campaign in the form of Public
Service Announcements on cyber-hygiene or some other
combination of efforts.
CONCLUSION
The Committee's review of the threat posed by internet
spoofing to our veterans has revealed that the issue is a
complex one, with ramifications extending well beyond the
veterans themselves to include their families, communities, and
ultimately the nation itself through attempts to influence our
elections. Spoofing has many manifestations, and through the
ubiquity of social media the potential reach of spoofers is
growing. As social media networks have expanded their reach and
diversified their platforms, new opportunities have arisen for
bad actors to leverage this technology to perpetrate scams and
misappropriate veterans' voices and images for malicious
purposes. Chairman Mark Takano noted, ``Manipulation of social
media networks, a major source of news and information, has
become a tool of influence.''\232\
---------------------------------------------------------------------------
\232\HVAC Committee Hearing Transcript, at 2.
---------------------------------------------------------------------------
Of importance is that a substantial number of these bad
actors are based in other countries, where some are even acting
at the behest of state-backed entities. Rep. Andy Barr (R KY)
stated it plainly by saying, ``We are very concerned about
scams and fraud schemes targeting our veterans coming from
overseas, foreign entities.''\233\ Foreign manipulation of
social media networks for the purposes of spreading
disinformation, sowing division, and influencing our political
elections is a clear, present, and growing threat facing the
veterans' community. This was an ongoing concern during the
2020 election year, as the lessons of foreign influence in the
2016 election still linger.
---------------------------------------------------------------------------
\233\Id. at 91.
---------------------------------------------------------------------------
Regrettably, the Committee also learned about significant
shortfalls in the efforts and abilities of the social media
platforms to detect, deter, and prevent such spoofing and
manipulation. There is no doubt that the major social media
platforms have developed global footprints which enable people
from all walks of life to quickly and easily connect with
friends and family, access news reports, financial services,
and commercial interactions. But the global reach and ease of
access also makes social media platforms particularly valuable
to spoofers seeking to efficiently, cheaply, and
surreptitiously encroach upon unsuspecting users, particularly
when veterans believe that they are interacting with a fellow
veteran.
Facebook and Twitter, two of the most significant and
influential social media platforms, testified before the
Committee and described the extensive resources that they are
devoting to studying, detecting, and deterring spoofers. Both
platforms noted the huge numbers of accounts they have closed,
the volume of content that has been removed, and their work
with outside analysts and cybersecurity experts to tighten
their platforms against future infiltration.
However, notwithstanding these efforts, Dr. Barash of
Graphika testified that the data shows steadily increasing
rates of spoofing--indicating that spoofers are outpacing the
platforms' efforts to curtail the problem. The platforms noted
specific challenges in addressing inauthentic behavior quickly
and comprehensively, including the need for multiple levels of
review to combine automatic detection of certain suspicious
activity patterns with subjective human review that can
accommodate protected usage like satire or tribute. Ultimately,
neither social media company was able to provide a definitive
answer as to how or when spoofing could be effectively
eliminated from their platforms.
Representatives of the FBI echoed the concerns about a
rapidly-evolving threat capable of learning how to circumvent
or defeat preventative measures put in place by the social
media firms, and described the difficulties in tracing spoofed
accounts back to individual bad actors or foreign networks
without closer collaboration between federal law enforcement
and the platforms. Drawing parallels to the existing reporting
requirements in the banking industry, the FBI articulated a
need for greater communication between the platforms and law
enforcement with a particular emphasis on early notification
that would allow the FBI and its partners to act proactively to
identify and even intercept bad actors before they are simply
able to reappear online under a different name, profile, or
guise. The proliferation of online threats and the presence of
foreign entities seeking to exploit veterans in order to pursue
illicit gains or disrupt American elections demands immediate
attention, and Congress can help bridge the existing gap
between the social media platforms and the law enforcement
community.
The law enforcement agencies largely rely on voluntary
disclosure of data by the platforms. The agency representatives
described the potential for protracted delays while pursuing
legal process in order to obtain specific data that the
platforms do not voluntarily share. Moreover, some of the FBI
representatives expressed concern that a system of mandatory
data sharing might be counter-productive if it resulted in the
platforms only sharing the specific mandated data in lieu of
the broader, more informal and transactional exchanges
currently in place. Therefore, options to expand the scope and
feasibility of voluntary data disclosures appear to be the
preferred course of action to improve law enforcement
efficiency. For example, the existing ECPA provisions for
voluntary disclosure of customer records and communications (18
U.S.C. Sec. 2702) already allow the platforms to provide
certain communications information to law enforcement under
narrowly defined circumstances in Sec. 2702(b), but stop short
of allowing voluntary disclosure of the type of identifying
account information that would be most useful in assessing and
apprehending potential criminals. The subsequent subpart,
Sec. 2702(c), which relates to the voluntary disclosure of
customer records (which would include the relevant identifying
account information) does not include corresponding language to
allow provision of these records to law enforcement. Aligning
the scope and nature of data available to law enforcement
through the voluntary disclosures authorized under the
different subsections of Title II of ECPA could be one way to
potentially enable the FBI to efficiently access the most
relevant data it needs for its enforcement work while
maintaining strong privacy protections and a voluntary
structure for information sharing.
Notably, there are relevant provisions compelling the
mandatory disclosure and sharing of certain protected financial
information in cases of terrorist financing, money laundering,
and other criminal conduct, which could serve as a model for a
more formal approach to data sharing with social media
platforms, should the voluntary disclosure approach prove
ineffective. While spoofing alone would not be likely to
trigger any criminal laws (short of commercial scams that could
constitute wire fraud), mandatory information sharing could be
facilitated by the agencies or required through a change in
statute. However, as noted by the law enforcement
representatives, there are significant reasons for preferring
voluntary disclosures, and so an expansion of such voluntary
disclosure authority should be the recommended approach.
Congress must consider the best way to facilitate a timely
exchange of detailed information that will enable the social
media platforms to honor their privacy commitments to their
users, while also positioning and equipping law enforcement
with the data it needs to identify and eliminate foreign actors
preying on unsuspecting Americans.
Moreover, Congress has an opportunity to help educate the
millions of users of social media, including veterans, about
the lurking threats posed by spoofers online. As Ranking Member
Roe observed, ``We want to shed light on the issues impacting
veterans, help them understand the risks associated with using
social media, and direct them to resources to empower them to
protect themselves and their families online.''\234\
---------------------------------------------------------------------------
\234\Id. at 8.
---------------------------------------------------------------------------
The data and analysis reviewed by the Committee demonstrate
that foreign attacks continue to proliferate, while it does not
appear that the platforms have implemented adequate
improvements to prevent spoofing preemptively. In conjunction
with greater alignment and communication between the social
media platforms and law enforcement, educating veterans and the
broader public about the threat of spoofing is an appropriate
and measured response to helping protect our veterans and
broader society.
As Chairman Takano laid out in the Committee hearing,
``Social medial platforms play an important role in public
discourse, and I continue to believe in protecting our freedoms
of speech and innovation. But there is a very real and growing
problem here, and we need to determine how to strike the
balance between shielding platforms from frivolous lawsuits and
ensuring election security and sanctity of our veterans''
voices in civic discourse. The platforms themselves need to do
more to eliminate the issue of internet spoofing, and if they
don't, then Congress will need to step in more directly.''\235\
---------------------------------------------------------------------------
\235\Id. at 5.
---------------------------------------------------------------------------
RECOMMENDATIONS
Recommendations and solutions to the threat of internet
spoofing fall into two broad categories. The first category is
oriented at users of social media and is defensive in nature,
such as teaching users how to be aware of the dangers posed by
spoofers on social media and training them how to protect
themselves through heightened vigilance, healthy skepticism,
and adherence to basic principles of cyber-hygiene. The second
category is aimed at putting the social media platforms and law
enforcement on the offensive and developing robust mechanisms
to more effectively identify and eliminate foreign-based
spoofers quickly. While the first category is likely to be less
costly and easier to implement, the second category may
ultimately prove to be more effective in bringing the threat
under control.
Improve Awareness
1. Improve Awareness through a Public Service Announcement
Campaign--As noted by several Committee Members, FBI
representatives, and testifying witnesses, the problem of
spoofing is exacerbated by a general lack of public awareness
of the issue and unfamiliarity with how to assess online
content in order to evaluate authenticity. Warnings of the risk
that social media content may not actually be from legitimate
sources or be deliberately planted for exploitative purposes
can be effectively and efficiently communicated through a
public awareness campaign, such as through public service
announcements (PSA). These public awareness campaigns can be
distributed through the social media platforms themselves, or
more comprehensively through other media outlets and agencies,
such as VA.
2. Develop Cyber-hygiene Training--VA and the Department of
Defense should develop robust and comprehensive cyber-hygiene
training. This would go beyond the basic information provided
by public awareness campaigns. For example, agencies could
provide training on best practices in protecting personal and
financial information, how to read and review content online
with an eye towards verification, and how to engage the
platforms themselves when needed to remove spoofed accounts,
fraudulent posts, or other deceptive content.
3. Strengthen Partnerships Between Social Media Platforms
and VSOs--A strong partnership could include an ongoing process
for VSOs to contribute their expertise and familiarity to
assist the social media platforms in their efforts to address
spoofing. The social media platforms noted that it can be
difficult to differentiate legitimate content from veterans or
VSOs from spoofed content purporting to be from the veterans'
community. There are ample resources within the broader
veterans' community to help advise and consult with the
platforms on such questions.
Strengthen Prevention and Enforcement Methods
4. Improve Reviews of Accounts by Social Media Platforms--
The social media platforms should implement stronger reviews of
accounts that pose substantial risk of spoofing. This should
include the adoption of industry-developed best practices
involving accounts that control groups or pages with very large
reach in order to closely scrutinize activity on these groups
or pages to quickly identify potential patterns of suspicious
behavior. Given the influence and reach, any such groups or
pages that meet or exceed certain thresholds of followership
should have their controlling accounts be officially verified
by the social media platforms, and the details of such
verification (ownership, geolocation, moderators, etc.) be
publicly available for all users.
5. Consider Legislative Reforms to Facilitate Sharing
Information--Congress should consider appropriate modifications
to the federal laws that currently limit the social media
platforms' ability to freely share data with law enforcement
agencies or other peer platforms in order to detect, prevent,
or remove fraudulent or spoofed content in a timely and
efficient manner. Federal law is murky on how the privacy
rights of users intersect with law enforcement needs with
respect to data or identification information in cases of
potential illegal activity or fraud. The platforms have
generally erred on the side of maintaining user privacy in the
absence of a clear legal requirement to provide such data to
law enforcement agencies. However, there are certain
inconsistencies in the existing laws governing voluntary
disclosures to law enforcement which contribute to challenges
and delays. Congress could align the scope of voluntary
disclosure of information to law enforcement under the
respective provisions of Title II of ECPA to facilitate greater
transparency and timely information sharing with law
enforcement. This would essentially allow holders of electronic
communications and records to voluntarily release the data
associated with fraudulent, spoofed, or misappropriated
accounts to law enforcement agencies and potentially also to
their enforcement counterparts at peer platforms, when criminal
activity or other imminent harm is reasonably suspected.
However, any new legislation in this area or any change to the
ECPA statute must be both narrow in scope and include strong
safeguards to protect the personal privacy and civil rights
concerns of users.
6. Increase Data Sharing on Fraudulent Accounts--Social
media platforms should improve their sharing of identified
fraudulent and spoofed accounts with other platforms and law
enforcement to the extent permissible under current statutes,
both in terms of frequency of sharing and the scope of the data
that is shared. Although ECPA protects underlying identifying
information, there is other information about spoofed accounts
that can still be shared. Increasing the scope and timeliness
of shared information pertaining to accounts that have been
identified, and likely removed as fraudulent or spoofed, would
enhance cross-platform detection. Additionally, consistent
protocols could be established around communication between the
platforms and law enforcement, and amongst the platforms, to
ensure that information is shared on a regular and timely
basis, rather than only in response to crises or incidents.
This sharing of information should be narrow in scope and
include strong safeguards to protect the personal privacy and
civil rights concerns of users.
7. Improve Identity Verification and Geolocation
Identification--Social media platforms should improve their
verification of identities, affiliations, and geolocation for
all accounts. This would create a consistent and more robust
version of the verification and checkmark system that was
previously employed in various permutations by Twitter and
Facebook. This would make it more difficult for foreign actors
to disguise or misrepresent their locations and consequently
their identities). The geolocation and account ownership
information should then be readily available to users and to
law enforcement, to increase transparency and foreclose
intentional concealment of where an account is based.
DISSENTING VIEWS
[F]oreign nations continue to use influence measures
in social and traditional media in an effort to sway
U.S. voters' preferences and perspectives, to shift
U.S. policies, to increase discord and to undermine
confidence in our democratic process. . . The American
public has a role to play in securing the election,
particularly in maintaining vigilance against foreign
influence. At the most basic level, we encourage
Americans to consume information with a critical eye,
check out sources before reposting or spreading
messages, practice good cyber hygiene and media
literacy, and report suspicious election-related
activity to authorities.\236\
---------------------------------------------------------------------------
\236\See https://www.dni.gov/index.php/newsroom/press-releases/
item/2135-statement-by-ncsc-
director-william-evanina-100-days-until-election-2020.
---------------------------------------------------------------------------
--William Evanina, Director, National Counterintelligence and
Security Center, Office of the Director of National Intelligence.
It is well recognized that the threat of foreign influence
on United States elections and disinformation campaigns using
social media is real. Moreover, the evidence demonstrates that
all users, including veterans, are targets of such
disinformation campaigns. The evidence also demonstrates that
the veteran community is targeted by criminals using social
media platforms to carry out fraud schemes.
Democrats and Republicans alike have expressed frustration
and anger towards the social media platforms, albeit, at times,
for different reasons. While we largely agree with the
Majority's recommendations to improve awareness and
communication between the social media platforms and the
veteran community, most of the Majority's recommendations in
this report go well beyond the Committee's jurisdiction and
expertise. The Majority wades into issues regarding how the
platforms should operate, suggests changes to federal data
privacy laws, and what information should be available to law
enforcement. These issues are not the providence of the
Committee on Veterans' Affairs and are being considered by
other committees. We are concerned that some of the
recommendations have constitutional and privacy implications
that the Majority has not fully considered. Finally, by
entering debates on how to address spoofing and disinformation
on these platforms, we move away from the Committee's core
focus of improving the quality of VA healthcare, benefits, and
services.
While we appreciate the Majority's efforts to educate the
veteran community on the disinformation and fraud threats they
face while using social media, for the reasons outlined below,
we must depart from the Majority on this report.
THIS IS NOT A BIPARTISAN REPORT
As an initial matter, the Majority's report was conceived
as and has remained a Majority product. We expect that before
announcing a bipartisan investigation, the Majority and
Minority staffs would discuss the objectives and scope of the
investigation. We also expect that a bipartisan investigation
would include an invitation to participate in all witness
interviews. Finally, we expect that the parties would discuss
the findings of the investigation and the objective of the
report before drafting it. Unfortunately, none of those things
occurred.
On March 5, 2019, Chairman Takano issued a press release
announcing the investigation. Minority staff was informed of
the investigation the day it was announced but was not involved
in the discussions about the investigation's objectives or
scope prior to the announcement. The Majority invited Minority
staff to participate in a meeting with the Vietnam Veterans of
America (VVA) to review its report. Minority staff raised
concerns over the Committee's lack of jurisdiction over several
findings and recommendations in VVA's report.\237\ Minority
staff participated in meetings with Facebook to discuss some of
the concerns raised by VVA. Both staffs worked with Facebook to
address VVA's concerns and helped facilitate Facebook's
verification of additional Veteran Service Organizations (VSO)
Facebook pages. The Committee held a hearing on the topic in
November 2019 and held a briefing with law enforcement in
January 2020. The report cites to witness interviews with
Twitter on September 10, 2019, and October 17, 2019, and with
Graphika on September 25, 2019, interviews in which the
Minority was not invited to participate.\238\
---------------------------------------------------------------------------
\237\In April of 2018, VVA raised its findings regarding veterans
being targeted on social media prior to the House Energy and Commerce
Committee's April 11, 2018 hearing on Facebook's transparency and use
of consumer data. See http://docs.house.gov/meetings/IF/IF00/20180411/
108090/HHRG-115-IF00-20180411-SD014.pdf.
\238\The report identifies interviews with Twitter on September 10,
2019, and October 17, 2019 (see FNs 203 & 207). The report identifies
interview with Graphika on November 5, 2019 (see FN 137).
---------------------------------------------------------------------------
The Minority was not consulted in the outlining or drafting
of the report. Rather, the Majority presented its report in
August and solicited Minority staff comments. After sharing
concerns with the Majority, the Majority said that the report
would be released informally as a Majority staff report. That
plan changed in October 2020, when the Majority advised us that
the Committee would be voting to mark-up the report in late
November or early December. The Majority shared the final draft
with all committee members on November 19, 2020, just days
before members went home for the Thanksgiving holiday. The
Majority's report was conceived as and has remained a Majority
product.
THE RECOMMENDATIONS POSE A POTENTIAL THREAT TO FREE SPEECH
We agree with the Majority that the United States must
combat foreign interference in our elections. The Majority's
report, however, unnecessarily inserts the Committee into
political and policy debates that are beyond the jurisdiction
and expertise of this Committee. Moreover, we are concerned
that the recommendations, in an effort to combat spoofing and
misinformation, may undermine free speech.
On October 6, 2020, Republican members of the House
Judiciary Committee issued a staff report detailing the
silencing of conservative viewpoints by Big Tech companies,
including the social media platforms (hereinafter, ``Censorship
Report'').\239\ The Censorship Report cited examples where
conservative voices, including the voices of Members of
Congress, were silenced. We are concerned that the Majority's
report and recommendations could further fuel the silencing of
conservative voices on social media platforms.
---------------------------------------------------------------------------
\239\https://republicans-judiciary.house.gov/wp-content/uploads/
2020/10/2020-10-06-Reining-in-Big-Techs-Censorship-of-
Conservatives.pdf.
---------------------------------------------------------------------------
For instance, the Majority draws on examples of content
associated with conservative views to demonstrate how spoofers
use social media to push ``political propaganda & socially
divisive content'' to drive engagement or push disinformation.
In that section, the Majority uses memes--one from ``Vets for
Trump'' and another from ``Vietnam Veterans''--that juxtapose
Colin Kaepernick, the former NFL player who kneeled during the
National anthem in protest of police brutality and racial
disparities, and former NFL players Pat Tillman and Glen Coffee
who left the NFL to join the Army, with Pat Tillman making the
ultimate sacrifice. The Majority states that, ``These are
examples of socially divisive images being used to place
veterans and `heroes'' on one side and those protesting police
brutality, or supporting Black Lives Matter, on the other.''
Colin Kaepernick started kneeling during the National
Anthem on September 1, 2016.\240\ President Obama defended Mr.
Kaepernick's protest on September 5, 2016.\241\ On September
12, then presidential candidate Donald Trump weighed into the
debate and voiced his opposition to the protest.\242\ The
kneeling debate is legitimate and even the veteran community is
divided over the issue.243}244 Accepting that the
examples used in the report were posted by non-U.S. actors
between March and August 2019, the debate on kneeling during
the National Anthem predated the posting by the foreign
actors.\245\ Similar messages appear to have been posted
outside the period the legitimate owners lost control. Thus, we
are concerned with the Majority's characterization of this
debate as ``political propaganda and socially divisive
content'' with the implication being that it should be
moderated to address spoofing could unintentionally undermine
free speech.
---------------------------------------------------------------------------
\240\https://www.sportingnews.com/us/nfl/news/colin-kaepernick-
kneeling-protest-timeline/xktu6ka4diva1s5jxaylrcsse.
\241\Id.
\242\Id.
\243\https://www.bostonglobe.com/metro/2017/09/25/military/
MLjykBWSTbMlWqbaVUAGzI/story.html.
\244\https://www.tennessean.com/story/news/politics/2017/09/28/
kneeling-anthem-veterans-
opinions-nfl/709488001/.
\245\According to a news report, the ``Vets for Trump'' Facebook
page was started in 2015 by two Americans, including a Navy veteran.
Individuals from North Macedonia highjacked the page between March and
August 2019. See: https://www.washingtonpost.com/technology/2019/09/17/
popular-facebook-page-vets-trump-seemed-be-place-former-military-
months-macedonians-
controlled-it/. However, the content of the ``Vets for Trump'' page
outside the six months the page was highjacked is consistent with
material that the Majority labels as ``divisive.''
---------------------------------------------------------------------------
Fear of censorship by the social media platforms is not
imagined. On October 14, 2020, Facebook and Twitter limited the
sharing of New York Post articles that alleged corruption by
Hunter Biden, the son of Democratic presidential nominee Joe
Biden.\246\ In response, the Senate Committee on Commerce,
Science, and Transportation held a hearing on October 28, 2020,
to examine Section 230 of the Communication Decency Act and
whether it needs modification.\247\ Political drama ran high
during the hearing with a Republican senator chastising the
social media platform representatives while a Democrat senator
claimed the hearing was ``a sham.''\248\
---------------------------------------------------------------------------
\246\https://www.wsj.com/articles/facebook-twitter-limit-sharing-
of-new-york-post-articles-that-biden-disputes-11602736535.
\247\https://www.commerce.senate.gov/2020/10/does-section-230-s-
sweeping-immunity-enable-big-tech-bad-behavior.
\248\https://www.cnn.com/2020/10/28/tech/section-230-senate-
hearing-wednesday/index.html.
---------------------------------------------------------------------------
On November 17, 2020, the Senate Judiciary Committee held a
similar hearing titled, Breaking the News: Censorship,
Suppression, and the 2020 Election, to examine censorship by
social media platforms and whether Congress needs to modify
Section 230 of the Communication Decency Act.\249\
---------------------------------------------------------------------------
\249\https://www.judiciary.senate.gov/meetings/breaking-the-news-
censorship-suppression-and-the-2020-election.
---------------------------------------------------------------------------
Another example from the report that poses a potential
threat to free speech can be found in the Majority's commentary
on Twitter's October 2019 decision to stop all paid political
advertising. In that section the Majority states:
Twitter's decision to ban paid political
advertisements has been roundly commended. However,
there are still loopholes which facilitate the
promotion of political agendas without conflicting with
the ad ban. Messages can be crafted around political
issues without naming specific candidates, parties, or
outcomes.
In the next section titled, ``Is Twitter Doing Enough?''
the Majority identifies one of the perceived loopholes.
Specifically, the report states:
Twitter continues to monitor and enforce political
accounts for compliance with its Rules on Platform
Manipulation. This was recently enforced against the
campaign of Michael Bloomberg in the Democratic
primaries, resulting in the suspension of seventy
accounts for coordinated behavior. However,
opportunities remain for spoofers to exploit gaps
between these policies, for example, by using divisive
content that does not meet the threshold of paid
political advertising, but serves similar purposes. For
example, Twitter still allows ads related to social
causes such as climate change, gun control, and
abortion, but organizations cannot advocate for or
against a specific political, judicial, legislative, or
regulatory outcome related to those matters.
(emphasis added)
Implicit in the Majority's commentary is that Twitter is
not doing enough to prevent what the Majority considers to be
divisive content from being posted to the platforms. The
Majority's solution appears to be to ban all paid advertising
on social media that touches upon any matter of political
consequence that someone could find divisive regardless of
where it originates. This is concerning because not all ads
related to social causes are placed by spoofers. Thus, the
impact could be that legitimate advocacy of conservative
positions is stifled. The silencing of voices is consistent
with the findings of the Censorship Report. Specifically, the
Censorship Report found that, ``When there is conflict over
contentious issues, the trend is to silence and suppress
disfavored views rather than address reasonable arguments that
best enrich America's civic discourse.''\250\
---------------------------------------------------------------------------
\250\https://republicans-judiciary.house.gov/wp-content/uploads/
2020/10/2020-10-06-Reining-in-Big-Techs-Censorship-of-Conservatives.pdf
at 1.
---------------------------------------------------------------------------
In addition, the Majority's report highlights problems with
the social media platforms' fact checking operations which
further amplify the risk of content moderation based on
viewpoint. In its analysis of Facebook under the heading, ``Is
Facebook Doing Enough?'' the Majority states:
Facebook has continued to draw attention and a
measure of criticism for its decisions to allow certain
doctored content on its platform that some users decry
as deliberately misleading or fake news. Compounding
the problem, Facebook partners with third-party fact-
checkers to assess the veracity of content and identify
misinformation, but defers substantially to the
discretion of those external parties as to what content
is actually fact-checked. Thus, even content that
Facebook asserts is ``eligible'' to be fact-checked may
not in actuality be examined unless a third-party
partner specifically selects that content for review.
The practical implication of this structure is that
Facebook is able to shift accountability away from
itself by pointing to its external fact-checking
partners, but then it does not appear to provide
sufficient guidelines on what content those partners
must review--thereby significantly eroding the efficacy
of its fact checking operations.
These conditions, in our opinion, are ripe for fostering
censorship.
We believe in the free exchange of ideas, regardless of
political ideology, even if those ideas make some
uncomfortable. We believe that by framing the issue of social
media spoofing and disinformation as a veterans' issue the
Majority inserts the Committee into a political debate for
which it has no jurisdiction. Given recent examples of the
silencing of conservative voices, as well as concerns regarding
content moderation, it is important to inform veterans of the
real threat of social media disinformation campaigns without
potentially undermining legitimate free speech.
WE CANNOT SUPPORT FIVE OF THE SEVEN RECOMMENDATIONS
The Majority offers seven recommendations broken into two
categories: (1) Improving Awareness, and (2) Strengthened
Prevention and Enforcement Methods. While we support
recommendations one and three, we respectfully disagree with
the Majority on the remaining recommendations.
Recommendation 1--Improve Awareness Through a Public Service
Announcement Campaign
Veterans, like all social media users, could benefit from a
campaign to improve public awareness about how to protect
themselves online. As the report notes, ``The FBI
representatives mentioned that one of the surest ways to limit
the reach of spoofers is to improve cyber-hygiene, or the
practices and precautions that users of computers and social
media should take to maintain system health and improve online
security.'' To that end we support Recommendation 1, which
would improve awareness of the risks associated with the use of
social media platforms through the social media platforms,
media outlets, and federal agencies.
The veteran community has a role to play in addressing this
issue as well. We commend Vietnam Veterans of America on its
recent efforts to partner with the Department of Homeland
Security's Cybersecurity and Infrastructure Security Agency, to
develop and promote cyber-hygiene materials to the veteran
community.\251\\252\
---------------------------------------------------------------------------
\251\https://www.cisa.gov/news/2020/10/23/cisa-and-vietnam-
veterans-america-partner-raise-awareness-about-threat.
\252\https://vva.org/protect2020/.
---------------------------------------------------------------------------
Recommendation 2--Develop Cyber-Hygiene Training
Recommendation 2 states that ``VA and Department of Defense
should develop robust and comprehensive cyber-hygiene
training'' that goes beyond basic information provided by a
PSA. We are concerned that VA may lack the expertise and
resources to ``develop robust and comprehensive cyber-hygiene
training'' which is not currently part of VA's mission.
While it is unclear how this recommendation would be turned
into action, we note that Chairman Takano offered an amendment
to the House Fiscal Year 2021 National Defense Authorization
Act which would establish within the Department of Veterans
Affairs a new Office of Cyber Engagement to develop and provide
cyber-hygiene services to veterans.\253\ This new office would
be headed by a member appointed to the Senior Executive Service
and report directly to the Deputy Secretary or Secretary. The
Committee has no legislative record to evaluate whether VA has
the existing expertise and resources to stand up this new
office and create cyber hygiene training.
---------------------------------------------------------------------------
\253\See Section 1802 of H.R. 6395; https://www.congress.gov/bill/
116th-congress/house-bill/
6395 /.
---------------------------------------------------------------------------
Additionally, other federal agencies like the Department of
Homeland Security, the Department of Justice, and the Federal
Bureau of Investigations have roles in election security,
combating disinformation campaigns, or prosecuting fraud.\254\
These agencies have developed cyber-hygiene training
materials.\255\ Before recommending that VA develop its own
cyber-hygiene office and products, we believe the Committee
should evaluate how best to leverage existing expertise and
resources to address the goal of educating veterans on cyber
security risks.
---------------------------------------------------------------------------
\254\On March 19, 2019, a bipartisan group of members, including
two on the Veterans' Affairs Committee, sent FBI Director Chris Wray a
letter, as head of ``the federal law enforcement agency responsible for
criminal and counterintelligence investigations,'' requesting the FBI
to investigate VVA's findings. See: https://cisneros.house.gov/media/
press-releases/bipartisan-veterans-demand-investigation-suspicious-vso-
social-media-accounts.
\255\https://www.fbi.gov/investigate/counterintelligence/foreign-
influence/protected-voices.
---------------------------------------------------------------------------
Recommendation 3--Strengthen Partnership Between Social Media Platforms
and VSOs
We support Recommendation 3, which would encourage the
social media platforms to partner with VSOs to address the
concerns of the veteran community. In early 2019, following
meetings with the Majority and Minority staff, Facebook worked
with the Committee to verify VSO Facebook pages to address very
real concerns that fraudsters were using fake VSO pages to sell
merchandise or disseminate information. Although we were not
included in the Majority's interviews with Twitter, the report
indicates that ``Twitter has informed HVAC that all VSOs with
Twitter accounts have now been verified, and has committed to
working with the committee to ensure that congressionally-
chartered VSOs, and their affiliated chapters, continue to be
verified.'' These examples demonstrate that by enhancing
communication and partnerships between the social media
platforms and veteran community, opportunities are created to
understand and address the VSO community's concerns.
Recommendation 4--Improve Reviews of Accounts by Social Media Platforms
Recommendation 4 calls for the social media platforms to
``implement stronger reviews of accounts that pose substantial
risk of spoofing,'' the adoption of industry-developed best
practices involving accounts that control groups or pages with
high follower threshold, and official verification of such
groups or pages with ownership and geolocation information made
publicly available for all users.
As the Majority's report notes, Facebook already implements
``higher standards of verification, visibility, and
transparency for pages that exceed a threshold for large
numbers of followers, political advertisers, and certain
commercial pages.'' The report then adds that Facebook private
groups are one way that inauthentic users avoid transparency.
Recognizing that Facebook already does what the Majority is
advocating for to an extent, we have privacy concerns related
to the proposed expansion.
First, this recommendation does not differentiate between
U.S. citizens or foreign actors. We are concerned that there
may be U.S.-based accounts with a large following that could be
captured by this recommendation. As discussed in our response
to Recommendation 7, we are concerned that the call to verify
all accounts over a defined threshold of followers, collect
geolocation information, and make the information publicly
available for all users could undermine free speech and raises
privacy concerns. We share the desire to protect and empower
the veteran community, but we do not believe this Committee has
the jurisdiction or expertise to wrestle with these complex and
sensitive issues.
Recommendation 5--Consider Legislative Reforms to Facilitate Sharing
Information
Recommendation 5 suggests that Congress should ``consider
appropriate modifications to the federal laws that the social
media platforms currently construe as limiting their ability to
freely share data with law enforcement agencies or other peer
platforms in order to detect, prevent, or remove fraudulent or
spoofed content.'' The Majority specifically calls for an
amendment to the Electronic Communications Protection Act
(ECPA) to facilitate the social media platforms' sharing of
information with law enforcement.
The jurisdiction of House committees is established by the
Rules of the House.\256\ The Committee on Veterans' Affairs'
jurisdiction broadly concerns veterans matters. The House
Judiciary Committee was the committee of jurisdiction that
reported the Electronic Communications Protection Act of 1986
(P.L 99-508).\257\ As such, the House Committee on Veterans'
Affairs lacks the jurisdiction and expertise in the nuances and
operation of the ECPA and federal laws regulating data sharing
by social media platforms.
---------------------------------------------------------------------------
\256\See Rules of the House of Representatives, 116th Congress,
Rule X, Clause 1(s).
\257\See https://www.congress.gov/bill/99th-congress/house-bill/
4952.
---------------------------------------------------------------------------
We recognize that members of this Committee may have ideas
on how to improve the ECPA and other data sharing laws
affecting the social media platforms. However, ``appropriate
modifications''' is vague. We note that the report refers to
the ECPA as an ``impediment'' to combating spoofing, but we are
unwilling to join our colleagues in recommending that a change
to the ECPA is appropriate because this is outside our
expertise. Rather, we defer to our colleagues in the committee
of jurisdiction who have the expertise to evaluate proposed
changes to ECPA.
Recommendation 6--Increase Data Sharing on Fraudulent Accounts
Recommendation 6 suggests that social media platforms
improve their sharing of fraudulent and spoofed accounts to the
extent permissible under current statutes, and that they
develop more consistent protocols regarding sharing this
information with other platforms and law enforcement should be
established.
We appreciate the Majority's recognition that
Recommendation 6 comes with privacy and civil rights concerns
and we echo those concerns. While we agree on the need to
protect our elections from foreign interference and veterans
from spoofing, there are important privacy and civil rights
concerns at issue in this recommendation. Therefore, we
disagree with the Majority's recommendation and defer to
expertise of the committee of jurisdiction.
Recommendation 7--Improve Identity Verification and Geolocation
Identification
Recommendation 7 suggests that social media platforms
should ``improve their verification of identities,
affiliations, and geolocation for all accounts''' and make this
readily available to users and law enforcement.
With an average of over 2 billion daily users across the
Facebook's platforms alone, the breadth of this recommendation
is extensive. Moreover, the Majority's recommendation makes no
distinction between a U.S. citizen or a foreign actor and calls
for sensitive information to be collected and made readily
available for all users and law enforcement. The Majority notes
that some platforms, like Facebook, have already taken steps to
verify certain accounts and make that geolocation information
available publicly. We are unprepared, however, to join the
Majority in this wholesale approach out of concern of its
impact on privacy and free speech of Americans.
The tradition of anonymous political debate is as old as
the United States. Courts have recognized the value of
anonymity in political debate. We are concerned with the
potential impact of the recommendations. For example, if the
social media platforms were to publish the identity,
geolocation, and affiliations information of all accounts--
including the accounts of American citizens engaging in
protected debate--such publication could undermine free speech.
Moreover, we are concerned about the privacy implications of
making this information available to law enforcement.
Assessing whether data privacy laws need to be updated or
whether the tools and information available to law enforcement
are adequate to address the realities of social media are
important matters that are currently being debated by the
committees of jurisdiction. They are not, however, matters
within the expertise or jurisdiction of this Committee.
CONCLUSION
Foreign adversaries like Russia, China, and Iran use social
media to sow division. They target many groups including the
veteran community. Scammers also use social media to perpetrate
fraud against the veteran community. We support the Majority's
recommendations to improve awareness of these issues among the
veteran community and foster cooperation between the social
media platforms and the veteran community. However, we do not
believe that a single hearing, a non-transcribed brief from DOJ
and the FBI, and staff interviews with the social media
platforms, some of which were not bipartisan, gives the
Committee on Veterans' Affairs the jurisdiction or expertise
necessary to wade into these complex and sensitive issues.
Therefore, we dissent to the Majority's report.
For the Minority,
David P. Roe, M.D.,
Ranking Member.