(WASHINGTON) — A report compiled by private researchers and expected to be released Monday by the Senate intelligence committee says that “active and ongoing” Russian interference operations still exist on social media platforms, and that the Russian operation discovered after the 2016 presidential election was much broader than once thought.
The report was compiled by the cybersecurity firm New Knowledge with data provided by the Senate committee from major tech companies Facebook, Twitter and Alphabet, the parent company of Google. Along with another report expected to be released by the panel, it is the first comprehensive analysis of the Russian interference on social media beyond what the companies themselves have said.
The report says that there are still some live accounts tied to the original Internet Research Agency, which was named in an indictment from special counsel Robert Mueller in February for an expansive social media campaign intended to influence the 2016 presidential election. Some of these accounts have a presence on smaller platforms as the major companies have tried to clean up after the Russian activity was discovered.
“With at least some of the Russian government’s goals achieved in the face of little diplomatic or other pushback, it appears likely that the United States will continue to face Russian interference for the foreseeable future,” the researchers wrote.
The report says that none of the companies turned over complete data sets to Congress and some of them “may have misrepresented or evaded” in testimony about the interference by either intentionally or unintentionally downplaying the scope of the problem.
One major takeaway of the study is the breadth of Russian interference that appeared on Instagram, which is owned by Facebook and was not as frequently mentioned when its parent company testified on Capitol Hill. The study says that as attention was focused on Facebook and Twitter in 2017, the Russians shifted much of their activity to Instagram.
The study says that there were 187 million engagements with users on Instagram, while there were 77 million on Facebook.
“Instagram was a significant front in the IRA’s influence operation, something that Facebook executives appear to have avoided mentioning in Congressional testimony,” the researchers wrote. They added that “our assessment is that Instagram is likely to be a key battleground on an ongoing basis.”
The Russian activity went far beyond the three tech companies that provided information, the report says, reaching many smaller sites as well. It details the sophisticated attempts to infiltrate internet games, browser extensions and music apps. The Russians even used social media to encourage users of the game Pokemon Go — which was at peak popularity in the months before the 2016 presidential election — to use politically divisive usernames, for example.
The report discusses even more unconventional ways that the Russian accounts attempted to connect with Americans and recruit assets, such as selling merchandise with certain messages, specific follower requests, job offers and even help lines that could encourage people to unknowingly disclose sensitive information that could be used against them.
The Russians’ attempts to influence Americans on social media first became widely public in the fall of 2017. Several months later, Mueller’s indictment laid out a vast, organized Russian effort to sway political opinion. While the social media companies had already detailed some of the efforts, the indictment tied real people to the operation and named 13 Russians responsible.
The study repeats several conclusions that Mueller, the intelligence community and the Senate and House intelligence committees had already made — that many of the postings focused on race, that they were primarily to hurt Democrat Hillary Clinton and help Donald Trump, and that the ultimate goal was to sow American division.
Also notable is the study’s finding that WikiLeaks founder Julian Assange was favorably treated in posts aimed at both left-leaning and right-leaning users. The report says there were a number of posts expressing support for Assange and Wikileaks, including several in October 2016 just before WikiLeaks released hacked emails from Hillary Clinton’s campaign.
The second report, compiled by researchers at the University of Oxford and called the Computational Propaganda Research Project, is also partly based on data from the Senate panel. It notes that peaks in Internet Research Agency advertising and organic activity — or posts, shares and comments by users — often corresponded with important dates in the U.S. calendar, crises and international events.
That report says the IRA’s posts focused on the United States started on Twitter as far back as 2013, and eventually evolved into the multi-platform strategy. The researchers from Oxford said that organic postings were much more far reaching than advertisements, despite Facebook’s sole focus on ads when the company first announced it had been compromised in 2017.
Other findings in the study:
- Russian activity on Twitter was less organized around themes like race or partisanship but more driven by local and current events and made use of occasional pop culture references;
- Facebook posts linked to the IRA “reveal a nuanced and deep knowledge of American culture, media, and influencers in each community the IRA targeted.” Certain memes appeared on pages targeted to younger people but not older people. “The IRA was fluent in American trolling culture,” the researchers say;
- Establishment figures of either party, especially Clinton, were universally panned. Even a tag targeted to feminists criticized Clinton and promoted her primary opponent Bernie Sanders;
- Several posts promoted the Russian agenda in Syria and Syrian President Bashar Assad;
- During the week of the presidential election, posts on right-leaning sites connected to the IRA aimed to generate anger and suspicion and hinted at voter fraud, while posts on sites targeted to African-Americans largely ignored mentions of the election until the last minute.
