As U.S. President Donald Trump and his allies flooded social media on Wednesday with false claims of victory and unsupported allegations of voter fraud, social media companies warned users the presidential election had yet to be decided.
The posts on Twitter TWTR.N, Facebook FB.O and other sites are imposing a real-time test of Silicon Valley’s much-touted rules on handling election misinformation and premature claims of victory.
But it is not clear whether the disclaimers and fact-checks, which often occur well after posts have been shared tens of thousands of times, are curbing the circulation of baseless claims. And in the case of Facebook, staff rewrote some of their rules on the fly.
Critics say that, in environments explicitly built to promote instant sharing and viral posts, bland disclaimers do not cut it.
“We’re on the brink here,” said Jessica Gonzalez, co-CEO of the advocacy group Free Press. She said Facebook should take down disinformation rather than just flag it.
“We’re testing our democracy experiment, and Facebook hasn’t gotten it right up to now,” Gonzalez said.
The false and unfounded claims evolved over the course of the day. Initially, Trump said he had won (he hasn’t.) Then he said that unanticipated mail-in votes were appearing out of nowhere (in fact they were long-expected.) Later, Trump’s campaign claimed it had won Pennsylvania (where votes are still being counted.)
Unsubstantiated allegations that ballots for Trump were not counted because voters used Sharpie pens, dubbed “Sharpiegate” on social media, were labeled as false on Facebook but shared hundreds of thousands of times.
The rumors prompted more than 100 Trump supporters, urged on by right-wing Twitter figures, to gather in Phoenix on Wednesday and demand to be let into an election building where votes were being counted.
Facebook’s rules evolved too, with eligibility for premature victory labels expanded to non-politicians based on “potential for virality” and shifting guidance on whether false claims of state victories would get labels.
The claims came as Trump’s lead in battleground states has evaporated as more ballots are counted in Michigan and Wisconsin, a process that has taken longer than usual as a result of the COVID-19 pandemic and extraordinarily high turnout. Trump’s Democratic challenger Joe Biden is projected to win Wisconsin and Michigan.
The extended count is no surprise, and neither was the late swing to Biden, which was widely predicted in the run-up to the vote – including by Reuters. That did not stop Trump and his supporters from flooding the internet with unfounded allegations of fraud.
Twitter marked a series of Trump’s tweets, including one making a false claim about “surprise ballot dumps,” as potentially “misleading about an election” and restricted sharing of the posts.
It also tagged statements by Trump’s son Eric and White House Press Secretary Kayleigh McEnany that falsely claimed victory in Pennsylvania.
Facebook put labels underneath several of the president’s false posts to note that “final results may be different from the initial vote counts,” but did not restrict interactions with the posts. Each was shared tens of thousands of times.
YouTube, the video service of Google parent Alphabet Inc GOOGL.O, likewise placed “results may not be final” labels below videos containing unsubstantiated claims without identifying the content as false or restricting comments.
NOT STOPPING THE SPREAD
Even if the disclaimers are curbing the spread of misinformation on their platforms, they have not stopped other media from amplifying Trump’s comments or his claims from hopping from one platform to another.
On Fox News, Trump’s tweets were read out verbatim on Wednesday, sometimes without caveats about their veracity. And on the video-sharing site TikTok, a group called the Republican Hype House shared a video featuring a false claim that Michigan found 138,000 ballots in a lake.
TikTok said the video was later removed for violating its policy against misleading information. Fox did not respond to an email seeking comment.
Alex Stamos, director of the Stanford Internet Observatory, told reporters one critical issue was the tender treatment accorded prominent figures who lie repeatedly on social platforms.
“A lot of it is the elite,” he said. “You see it over and over again.”
Social media companies have been under scrutiny over how they police rapidly spreading false information and election-related abuses of their platforms. In the weeks before Tuesday’s vote, both vowed action on posts by candidates trying to declare early victory.
The attempts at policing those declarations began early on Wednesday, when Twitter hid a Trump tweet that claimed “we are up BIG, but they are trying to STEAL the Election” behind a label that said it was potentially misleading.
Facebook added a label to the same post, which had about 25,000 shares.