A tweet from liberal activists touting Elizabeth Warren drew what seemed like a typical response from one of the Democratic presidential candidate’s fans this September:
“Thank you for endorsing Elizabeth Warren!!!” the user wrote, sharing a photo of black women holding “African Americans with Warren” signs.
The post gained only a single retweet at the time. But it found new life this past weekend, making its way to sharp-eyed Twitter users who realized it was fake, with the campaign placards photoshopped over “Black Lives Matter” signs.
Twitter users seized on a side-by-side comparison of the doctored version and the original, assailing the Warren campaign for the apparent misrepresentation. What they did not realize was that the account that had propagated the photo has been identified by the Warren campaign as a “troll,” only feigning support for the Massachusetts Democrat as it pushed out falsified content in an apparent effort to undermine her candidacy.
As the image solidified negative views of Warren among some already favoring other Democratic candidates, the incident offered a fresh lesson about political disinformation: Homespun operations on social media represent a rising threat, capable of inciting conflict among voters and turning unwitting users into agents of online deception.
Four years after Russian agents weaponized social media during the 2016 election, tech giants are grappling not just with foreign meddling but with falsehoods spread by less sophisticated, and frequently U.S.-based, online sources. Such actors already have circulated misleading posts, doctored photos and manipulated video around the 2020 race.
The threat is especially acute as Twitter and its Silicon Valley peers maintain a mostly hands-off approach to deceptively edited content, so long as the originating account isn’t engaging in behavior designed to “artificially amplify or suppress information,” as Twitter’s rules provide.
Relying on the anonymity and amplification that social media offers, such subterfuge “doesn’t necessarily change minds, but it certainly pushes us farther apart and it entrenches us in our existing positions,” said Darren Linvill, an associate professor at Clemson University who studies disinformation. “With these little home-grown cases that are clearly fake, and that a reasonable person can observe to be fake, someone who is already inclined to believe that thing is going to believe it.”
Social-media giants say they are not arbiters of truth, and in the past, Twitter has argued that the conversation around misleading or incorrect tweets can surface much-needed context for users without requiring the company to intervene.
But 2020 Democratic candidates have warned that they lack the tools to counter disinformation in real time. While the manipulated photo distributed by the suspect supporter account gained only moderate traction, its spread still has raised fears about what might happen if a more significant falsehood reverberated more widely.
In this case, early detection of the account responsible for the deceptive post had little benefit.
“Our campaign has reported this account, including this specific photo, several times before and Twitter has refused to take it down because, according to them, it does not violate their guidelines,” Chris Hayden, a spokesman for the Warren campaign, said, calling the Twitter persona a “troll account that has no affiliation with the Warren campaign.”
Twitter declined to answer specific questions about the account, including whether it was in fact operating from the United States. The account lists its location as Cobble Hill, Brooklyn, in New York. The user did not respond to a message sent over the site on Sunday.
A Twitter spokeswoman, Katie Rosborough, offered only a general statement on the episode. “All presidential campaigns have been offered the same level of support, including how to report issues,” she said.
It’s hardly the first time the platform has opted not to erase plainly false material designed to undermine a 2020 candidate. In September, Twitter said it would not scrub its platform of a conspiracy theory falsely tying Beto O’Rourke, the former Texas congressman who ended his presidential bid earlier this month, to a gunman who killed seven people in two west Texas towns. That decision drew a rebuke from O’Rourke’s campaign manager, Jen O’Malley Dillon, who sent letters to major tech firms imploring them to do more to root out disinformation. “This is your job, not ours,” she wrote.
The issue reaches beyond the presidential race, flaring up as well in other vexed partisan debates. Last spring, Twitter declined to remove a video of House Speaker Nancy Pelosi that made her appear drunk, even after the social media platform acknowledged it had been manipulated.
Over the weekend, a doctored photo of Rep. Adam Schiff, D-Calif., the chairman of the House Intelligence Committee and architect of the impeachment inquiry, spread on both Twitter and Facebook. In the altered version, the cropped head of Jeffrey Epstein, the convicted sex offender who died by suicide in August, was placed over the head of Schiff’s father, who had posed with his son for a photograph originally posted by the Democratic lawmaker last Thanksgiving. Some users took it upon themselves to post the word “fake” over the doctored photo, but others continued this week to share the false smear.
The Warren incident began on Sept. 30, when the Working Families Party sent a tweet touting the Massachusetts Democrat, whom the party had endorsed, as well as Sen. Bernie Sanders, I-Vt., also a candidate for the White House. Soon, a jubilant response arrived from @HelenManfred, an account with a small following that features what appears to be a manipulated image of a Warren look-alike as its header photo.
Using exaggerated language, the account predicted that working families, and people of color in particular, “are starting to SEE that she has a PLAN for them and they will GET ON BOARD!!!”
Included in the tweet was the doctored photo of the Black Lives Matter protest, which the account has since shared multiple times.
Over the weekend, the photo found its way to a supporter of Sen. Kamala Harris, one of Warren’s rivals for the Democratic nomination. The supporter, Leaundra Ross of El Paso, Texas, paired the doctored photo with the original and blasted it out to her nearly 5,000 followers. She called attention to the inclusion in the doctored photo of an image of “Charlamagne tha God,” the stage name of Lenard Larry McKelvey, a radio personality with more than 2 million Twitter followers and 126,000 YouTube subscribers. McKelvey famously confronted Warren during an interview about her claims of Native American ancestry.
“Y’all,” Ross wrote. “This photoshop is a mess.”
Within several hours, the tweet expressing Ross’s outrage had gained well over 1,000 interactions, including a string of comments assailing the Warren campaign for the apparent misrepresentation. Ross warned her followers that the photoshop was too poor to be an official production of the campaign. But some of her followers had already drawn their own conclusions. It was reposted by other Harris supporters, some of whom blamed Warren directly. “These are so much worse than her crazy lies about her plans,” one user wrote.
Meanwhile, Ross did not realize she had spread disinformation targeting a Democratic candidate.
“I posted it because it was funny and laughable and looked fake,” Ross said, speculating that the edited photo may have come from a well-meaning supporter when, in fact, a review of the account’s posts makes clear it is posing as a Warren fan to sow doubts about her campaign and spotlight apparent faults with her candidacy – as well as those of other 2020 hopefuls. “Warren 2020,” the user’s bio reads.
– – –
The Washington Post’s David Weigel contributed to this report.