Facebook sharply increased on Wednesday its estimate of how many Americans were exposed to Russian propaganda around the 2016 election after pointed and even angry questioning from the Senate intelligence committee.
Adding into account the reach of Russian propaganda on Instagram, which Facebook owns, Facebook vice president and general counsel Colin Stretch conceded that “gets you to a little less than 150 million” people.
It’s a far cry from company founder Mark Zuckerberg’s claim that it would be “crazy” to think that Russian trolls on Facebook had any election impact. And it was the latest revision from Facebook.
In early September, Facebook announced that it found 470 accounts linked to Russian propaganda pushing about 3,000 paid ads. On Monday, ahead of a gauntlet of Capitol Hill hearings, Facebook moved beyond the paid ads and revised its estimate to include organic, non-paid content, pushed by Russia and shared by unsuspecting Americans. The company came up with a new number: some 126 million Americans, far more than journalistic estimates.
On Wednesday, as Facebook and its tech colleagues Twitter and Google took heat from angry Senators investigating the scope of Russian electoral interference, the social media giant ratcheted its estimate up further, and indicated that it won’t be the last time. Senior attorneys for all three companies said their investigations of their platforms’ exploitation by the Kremlin and its allies is unfinished. CEOs of the companies did not attend, an absence bitterly noted by several legislators.
It was not the only stark number Facebook presented – again, for the first time – to bind the scope of the problem.
Stretch testified that Donald Trump and Hillary Clinton’s campaigns spent a combined $81 million on Facebook ads. By contrast, the Kremlin-tied troll farm known as the Internet Research Agency alone spent an estimated $46,000 on Facebook ads, speaking to a basic financial asymmetry Russia used as a lever into the 2016 election. Facebook had previously identified $100,000 worth of ads from the 470 Kremlin-tied accounts.
U.S. intelligence agencies have found Russian interference occurred – though President Donald Trump has called it a hoax – but did not assess whether it was decisive. (Under questioning, Stretch said, “the existence of those ads were on Facebook and were not a hoax.”)
Stretch indicated that “as early as 2015” Facebook was aware of “Russian state actors… engaged in more traditional cyber threat activity” like attempting to compromise accounts and shared it with law enforcement. (Stretch neglected to mention that Zuckerberg blew off warnings of Russian political interference in meetings with Facebook employees.) Twitter said “in 2015” it was aware of and took down accounts associated with the Internet Research Agency. Google’s Kent Walker said his company only conducted a “deeper dive” into the issue after the January issuance of a U.S. intelligence report on Russian electoral interference.
“I must say, I don’t think you get it,” said Sen. Dianne Feinstein, a Democrat who represents California, where all three companies are headquartered. Maine Independent Angus King said Russia had used their platforms for “a sophisticated worldwide strategy” to disrupt elections across Europe. Mark Warner, the senior Democrat on the intelligence committee, warned: “They are still doing it now.”
While all three company executives sought to assure legislators that they are seized with the propaganda problem, as they did on Tuesday to the Senate judiciary subcommittee, they seemed to offer outside actors a way to keep launching online disinformation campaigns into the foreseeable future. Twitter’s Sean Edgett indicated to Sen. Marco Rubio, a Florida Republican, that only by proxy – such as inflammatory posted content – would a foreign influence campaign be considered a violation of their terms of service. Facebook seemed to indicate that it was indeed a violation, while Google’s Kent Walker did not offer an answer.
Facebook was not the only company to come under criticism for understating the reach of Russian propaganda. Warner accused Twitter of “vastly underestimating” the extent of Russian penetration on its platform.
“Independent researchers have estimated that up to 15 percent of Twitter accounts, or potentially 48 million accounts, are fake or automated. Despite evidence of significant incursion and outreach from researchers, Twitter has, to date, only uncovered a small percentage of that activity,” Warner said, though he credited the company with recent expanded estimates.
Twitter’s Edgett, however, insisted that such bot activity is “less than five percent” of the platform, prompting the panel’s chairman, Richard Burr, to ask Twitter to formally explain the discrepancy.
Over the past month, legislators investigating Russian propaganda on social media have expressed deep dissatisfaction with Twitter for belated and apparently incomplete discoveries – or disclosures – over Russia hijacking the microblogging service. As of October 19, the extent of its disclosure to the committee was a batch of 1,800 promoted tweets from Kremlin network Russia Today, which paid the service $274,000. Twitter had said in late September that it had found and suspended 201 accounts connected to those suspended by Facebook for suspected Russian propaganda.
Shortly before this week’s hearings, Twitter suddenly discovered vastly more, suspending over 2,700 accounts linked to the Internet Research Agency that posted over 131,000 tweets in just two months around the 2016 election’s denouement. Right before the election, some 36,000 bots on Twitter aimed 1.4 million election-relevant tweets that received 288 million impressions from fellow tweeters.
“I doubt that the so-called Internet Research Agency represents the only Russian trolls out there,” Warner said, adding that Facebook “has more work to do to see how deep this goes.”
On Tuesday before the Judiciary subcommittee, Google, Facebook and Twitter executives prioritized a concern over “inauthenticity” on their platforms – Russian imposter accounts masquerading as Americans. Several senators were dissatisfied with what they considered an insufficient willingness on the part of the companies to grapple with, let alone police, the substance of the propaganda, which included Islamophobic, anti-refugee and pro-secession messages.
While Facebook said it is still investigating the extent to which the Kremlin exploited its platform around the 2016 U.S. election, it has been more definitive in closer to real time around similar Russian activity in European elections.
In April, on the eve of the French elections, Facebook suspended “over 30,000 fake accounts.” Last month, Facebook took additional action against an unspecified number of accounts that Richard Allan, its vice president of public policy for Europe, estimated totalled in the “tens of thousands” meddling in the German election.
Facebook has yet to acknowledge whether and to what degree Russian propaganda on its platform appeared during the United Kingdom’s 2016 debate over Brexit, the UK’s referendum on leaving the European Union, a foreign policy objective of Russia’s.
Warner pressed Facebook’s Stretch if the company had cross-checked the accounts it shuttered for interference in the French election for activity in the U.S. election. Stretch was less than definitive, prompting outrage from Warner, who indicated he had raised the question to Facebook in private meetings.
Shortly after their exchange, Facebook’s chief security officer, Alex Stamos, gave a different answer about its newer technological efforts to detect fake accounts, saying there was “not much overlap of activity in different countries from the same accounts.”
“After testing these techniques in the French and German elections we rolled them out globally, including in the U.S. By the time we discovered the cluster of accounts we assessed to belong to the Internet Research Agency this summer, around half had been disabled by our systems, demonstrating their effectiveness,” Stamos tweeted.
“As for Senator Warner’s specific question, all of the accounts disabled automatically are still included in our searches for organized disinformation actors like the Internet Research Agency.”