Statement for the record from Renee DiResta, Director of Research, New Knowledge
Honorable Committee Members –
My name is Renee DiResta, and I research influence operations and social network
manipulation. I appreciate the opportunity to submit this written and verbal testimony to
your committee.
Over the past decade, disinformation, misinformation, and social media hoaxes have
evolved from a nuisance into high-stakes information war. Our frameworks for dealing
with them, however, remain the same — we discuss counter-messaging and
counter-narratives, falling into the trap of treating this as a problem of false stories
rather than as an attack on our information ecosystem. We find ourselves in the midst of
an arms race, in which responsibility for the integrity of public discourse is largely in the
hands of private social platforms, and determined adversaries continually find new ways
to manipulate evolving feature sets and circumvent new security measures. It is critical
to acknowledge that computational propaganda and disinformation is not about
arbitrating truth, nor is it a question of free speech. Information Warfare is a
cybersecurity issue, it is an ongoing national security issue, and it must be addressed
through a collaboration between governments responsible for the safety of their citizens
and private industry responsible for the integrity of their products and platforms.
Propaganda and malign narratives have existed for a very long time, but today’s
influence operations, which co-opt popular social platforms, are materially different – the
propaganda is shared by our friends, often in the form of highly effective, shareable,
immediately graspable memes. It is efficiently amplified by algorithms, and the
campaigns achieve unprecedented scale. To conduct an operation, adversaries
leverage the entire media ecosystem to push a narrative and manufacture the
appearance of popular consensus. The operation is planned on one platform, such as a
messaging or chat board. Content is created, tested, and hosted on others, such as
Reddit, Pinterest and YouTube. It’s then pushed to platforms like Twitter and Facebook,
with standing audiences of hundreds of millions of people, and targeted at those most
likely to be receptive to it. The platform’s trending algorithms are gamed to make the
content go viral – this often delivers the added benefit of mainstream media coverage,
increasing attention via traditional media channels including television. If an operation is
successful and the content gets wide distribution, or a manipulative Page or Group
gains enough followers, the recommendation engine and search engine will continue to
serve up the content on an ongoing basis.
We are here because the Internet Research Agency (IRA) employed this playbook,
conducting an operation that leveraged our social networks to spread propaganda and
disinformation directly to American citizens. Their operation likely began sometime in
2013, continued throughout the 2016 election cycle, and even increased on Instagram
in 2017. While many accounts were shut down in 2017 as the tech companies began
their investigations, Twitter accounts and Facebook pages associated with the IRA
remain active. The IRA content on Facebook and Instagram alone had hundreds of
millions of engagements; Facebook itself estimates 146 million users across the two
platforms were affected. The Internet Research Agency’s disinformation campaign was
conducted on all the major platforms in the social network ecosystem. The presence of
manipulated content on Facebook and Twitter is well-documented. In the case of
Alphabet, YouTube, G+, and other properties were leveraged to either host content or to
support personas. Reddit, Tumblr, and Medium have confirmed that they were misused;
Twitter’s Vine video app was co-opted as well, and IRA meme boards were discovered
on Pinterest. Games and music apps were created and pushed to teenagers to
download. Even popular game Pokemon Go was incorporated into the operation.
Outside of social platforms, a number of websites were created to host original written
content, many of which looked very much like citizen journalism-style blogs and think
tanks. Topics ran the gamut, from social issues to concerns about war, the environment,
corporate greed, GMOs, energy policy, and immigration. Twitter accounts were created
to masquerade as local news stations. White House petitions were either created or
co-opted to engineer a perception of social consensus. Dozens of Facebook Events
were promoted, and activists were contacted personally via Messenger, to take the
operation to the streets.
The Internet Research Agency’s campaign pressed on a variety of socially divisive
issues, but the primary focus was on racial tension. Despite YouTube’s claim that the
content attributed to the IRA on its platform was “not targeted to the US or to any
particular sector of the US population”, it appears that the overwhelming majority of the
videos we are aware of were related to issues of importance to the black community,
particularly officer-involved shootings. Hundreds of thousands of Americans liked
Facebook Pages with names like Blacktivist, Heart of Texas, and Stop All Invaders. The
percentage of explicitly political content that mentioned candidates by name was small –
approximately 10% of known content – but the political content targeting both
right-leaning and left-leaning Americans was unified in its negativity toward the
candidacy of Secretary Clinton. In pages targeting the left, this included content
intended to depress voter turnout among black voters, or to paint Secretary Clinton in a
negative light as compared to candidates Jill Stein or Senator Bernie Sanders. Only the
social networks that hosted this campaign are in a position to gauge its full impact in
changing voter attitudes on their respective platforms. However, independent of its
impact, the fact that it was attempted, went undetected, and achieved such significant
reach is sufficient cause for concern.
Although this hearing was convened because of the Internet Research Agency’s
interference in the 2016 election, Russia was not the first to target American citizens
with propaganda using the social ecosystem. In 2014, the co-opting of social
communication infrastructure rose to mainstream awareness in the United States as
ISIS established a virtual caliphate, using every social app imaginable to push
propaganda boldly and transparently, using the features of our of social ecosystem in
precisely the way they were meant to be used: to build an audience and connect with
followers. This was a visible indication that the tools built to enable marketers and
messengers and friends to communicate could be co-opted and misused; the ensuing
debate about what to do about the problem made it apparent to anyone watching that
no one was in charge, and that American companies, American civil society
organizations, and the American government were deeply divided on how to respond to
the threat. That confusion continues even as the threat expands beyond extremists and
state actors: the Wall Street Journal recently revealed that a private intelligence
company, Psy-Group, openly marketed their ability to conduct similar types of influence
operations to impact the 2016 election.
As the internet has evolved, we’ve seen the consolidation of users into large standing
audiences on a small handful of social networks. This infrastructure has been a
phenomenal tool for small businesses to reach customers, and for the previously
voiceless to find a voice. But like any tool without appropriate safeguards, it can be
misused. These platforms employ gameable algorithms, and facilitate personalized
targeting that is enabled by the ongoing collection of extensive amounts of personal
data. As a result, social networks continue to be the most effective vector to manipulate
public sentiment and cause lasting damage to our democratic process. To combat this
evolving threat, we have to address those structural weaknesses and design an
effective deterrence strategy.
Individually, several social platforms have begun to take steps to reduce the spread of
disinformation, by disrupting economic incentive structures, reducing the spread of
clickbait headlines, and reducing the granularity of targeting criteria that were used to
push malicious content directly to subsets of the American people. Political ad content
on Facebook and Twitter is somewhat public now; we look forward to this database
being searchable via API, better equipping researchers and journalists to understand
our paid political conversation. Several platforms are beginning to take source quality
into account, which may help curb the ability of manipulative propaganda websites to
reach their audience. These steps, several of which were inspired by prior hearings in
this chamber, are a good start. But as platform features and protections change,
determined adversaries will develop new tactics.
In addition to the ongoing exploitation of social divisions, targeting of elections, and
disinformation about geopolitical events (such as the conflict in Syria), campaigns
targeting U.S. industry have emerged and are thriving. Influence operations are
increasingly appealing to a variety of actors: ideological true believers, non-state
extremists, economically-motivated enterprises, and State Actors. The last of these
requires a whole-of-government defense strategy, as it’s unlikely that commercial
platforms will be able to compete with the sophistication of well-resourced and
motivated hostile foreign government experienced in bypassing common security
checks.
The gaps in our ability to combat this type of information warfare became apparent
while attempting to address ISIS propaganda: the U.S. government was legally
constrained in its ability to respond, and the social platforms proved slow to act as
extremist content, assisted by platform targeting algorithms, easily made its way into the
social feeds of Americans. That wake-up call fell on deaf ears as our adversaries
prioritized, deployed, and perfected their influence operation capabilities. They were
able to exploit gaps in our intelligence community’s authorities, and take advantage of
our commitment to civil liberties; this left social platforms in the impossible position of
having to individually respond to this global threat, which has resulted in the
implementation of inadequate solutions and self-serving defensive policies on the part
of those private companies.
Several years after the threat emerged, the U.S. Government and the tech industry
respectively took small steps towards combating this threat by establishing the Global
Engagement Center and the Global Internet Forum to Counter Terrorism. The focus of
the latter is still solely terrorism, although the Global Engagement Center’s mandate has
expanded to countering foreign propaganda. The DOJ and NSA & Cyber Command’s
recent announcements that they will prioritize the mitigation and prosecution of this
activity is a positive sign. However, addressing this asymmetric threat requires a 21st
century Information Operations Doctrine, the implementation of a global real-time
detection and deterrence strategy, and the cooperation of private industry, press, law
enforcement, and the intelligence community.
The evolution of social media propaganda and influence techniques will bring serious
threats. We should anticipate an increase in the misuse of less popular and less
resourced social platforms, and an increase in the use of peer-to-peer messaging
services. We believe that future campaigns will be compounded by the employment of
witting or unwitting U.S. Persons through whom these state actors will filter their
propaganda, in order to circumvent detection by social platforms and law enforcement.
We should anticipate the incorporation of new technologies, such as videos and audio
produced by artificial intelligence, to supplement these operations, making it
increasingly difficult for citizens to trust their own eyes.
This will be one of the defining threats of our generation. Influence operations exploit
divisions in our society using vulnerabilities in our information ecosystem. They take
advantage of America’s commitment to freedom of speech and the free flow of ideas.
The social media platforms cannot, and should not, be the sole defenders of democracy
and public discourse. In that light, here are several recommendations we are proposing
toward achieving the goal of restoring integrity to the information ecosystem:
First, to address the most pressing short term issue, we recommend immediate
government action to identify and eliminate malign influence campaigns and to educate
the public in preparation for the 2018 elections.
Second, this domestic defense must be complimented by an updated global IO doctrine
and international detection and deterrence strategy, with the goal of mitigating foreign
influence targeting our allies, including the clear delegation of responsibility of this
activity within the U.S. Government. Taking example from the U.S. Government’s cyber
security response over the past decade, we must implement legislation that defines and
criminalizes foreign propaganda that targets not just our political process but also
addresses the targeting of commercial industry and social issues. Empowering law
enforcement with updated legal tools to investigate and prosecute sophisticated foreign
propaganda is essential to combatting this threat in the age of information warfare.
Third, the private tech platforms must be held accountable to ensure that they are doing
their utmost to manage and mitigate the pervasiveness of disinformation and
manipulative narratives in our privately-owned public squares. A number of regulatory
frameworks are on the table, including mandating that automated accounts be labeled,
limiting high-frequency advertising practices, and curtailing and reporting inauthentic
accounts. Regardless of which is chosen, and whether these policies are implemented
voluntarily by platforms (self-regulatory action) or via formal regulation, the incorporation
of oversight is key.
Finally, given that this asymmetric persistent threat impacts our social, geo-political, and
economic spheres, and given the sophistication of its tradecraft, we need new
structures for cooperation and information sharing between the public and private
sectors. Formal partnerships between security companies, researchers, and
government will be essential to defending our values, our democracy, and our society.
Thank you