Stephanie Baker and Priyanjana Bengani
London: Russian disinformation operators are deploying an army of bots to inflate view counts of fictitious posts about celebrities and Democratic candidates ahead of the US midterm elections.
AI-fabricated video clips of well-known actors and musicians – including Robert De Niro, Julia Roberts and Cher – attacking Democrats began appearing on X in early September, some urging people to join a campaign dubbed #AllDemocratsAreCriminals. Scores of other clips have made false allegations of corruption and improprieties by Democratic candidates.
Social media researchers have attributed the posts to a Russian state-backed operation known as Matryoshka, or Operation Overload, which floods social media with clips purportedly from established media outlets. Bloomberg News reviewed 130 videos and doctored newspaper front pages targeting the US midterms published by this network from September 10 to October 9, most of them flagged by anonymous disinformation research group Antibot4Navalny. US-based research groups Graphika and dTeam corroborated and supplemented the narratives.
One clip carrying an ABC News logo alleged that James Talarico, the Texas Democrat running for Senate, paid OnlyFans models. Another, with a CBS News logo, claimed the Internal Revenue Service had discovered Georgia Democratic senator Jon Ossoff owns eight undeclared properties. Talarico’s spokesperson has denied the candidate paid any OnlyFans models, and there’s no evidence the IRS discovered any undeclared Ossoff properties.
The Matryoshka videos appear to get tens or hundreds of thousands of views on X. But Antibot4Navalny researchers exclusively shared archived snapshots of posts showing how view counts are being manipulated by a network of automated accounts, called bots, to manufacture the illusion that many of the fabricated stories are popular.
Russia has used social media to influence US elections since 2016, when it sought to undermine confidence in the voting process and damage Hillary Clinton’s presidential campaign, according to US intelligence assessments. But its tactics have evolved over the past decade, from exploiting divisions across the political spectrum to focusing solely on attacking Democrats, who have been more supportive of Ukraine than Republicans.
Unlike previous elections, when it focused on presidential campaigns, Russia is trying to spread false stories about Democrats in numerous races nationwide, including 29 of the 36 running for governor and 20 of the 35 vying for Senate, according to Bloomberg’s review.
The US Office of the Director of National Intelligence shared a report with Congress in recent weeks warning that foreign adversaries were using artificial intelligence to spread disinformation online ahead of the midterm elections and that Russia was the most active in using digital tools to undermine confidence in election integrity, The Washington Post has reported. A spokesman for the Kremlin didn’t immediately respond to a request for comment.
AI has made it increasingly cheap and quick to produce fakes, allowing campaigns like Matryoshka to try to overwhelm social media and sow doubt about what can be trusted online. It’s the disinformation equivalent of drone warfare.
Antibot4Navalny, which has been tracking Russian bot activity since 2018, identified the influence campaign after noticing social media posts beginning in September 2023 that sought to undermine Western support for Ukraine and draw mainstream media into amplifying false claims. It named it Matryoshka after the Russian nesting dolls, because each day an account that tried to spread the previous day’s fake story posts a new one.
The research group, which started out monitoring disinformation about Russian opposition leader Alexey Navalny, who died in prison, did not disclose its location or any individual identities so it wasn’t targeted by Russian operatives, one member said in a message.
Russian users on Telegram first shared some of the fabricated content, according to a 2024 report by Viginum, the French government agency monitoring foreign digital interference, which cited Antibot4Navalny’s research.
“It’s evolving with the technology and is more sophisticated as a result,” said Darren Linvill, a professor at Clemson University in South Carolina and co-director of its Media Forensics Hub. “They’re actively trying to fish for more engagement.”
The number of views for Matryoshka posts can jump as high as 200,000 within the first hour but then quickly plateau, a phenomenon flagged by Antibot4Navalny researchers. Organic view counts usually follow a growth curve rather than a spike and abrupt flatlining in view counts. Most of the accounts that posted the fabricated videos have only a few followers and were dormant for years, Bloomberg found. Reposts and likes made up a tiny fraction of views.
Ukrainian President Volodymyr Zelensky said on October 12 that his country’s retaliatory strikes on Russian data centres had resulted in a drop in bot activity on social media, fuelling a story that has spread widely online.
But Graphika and AntiBot4Navalny say they’ve seen no evidence of a decline in Kremlin-linked bot activity on social media since the strikes started October 8. What declined, they said, was automated web-crawler traffic on Yandex, Russia’s main search engine. A spokesperson for Zelensky didn’t respond to a request for comment.
Russian operators pump out five to 10 narratives every weekday, often between 9am and noon, and take breaks on weekends. Most posts tag the social media accounts of mainstream news organisations. Matryoshka operatives have also bombarded fact-checking organisations with emails urging them to investigate the fake posts in an apparent bid to generate media coverage.
Matryoshka posts share common characteristics and follow a unique distribution pattern: fake video reports impersonating media outlets; long-dormant accounts suddenly repurposed to post one or two videos; tagging and emailing media organisations; and multiple posts on the same day around a single theme.
Most videos run for roughly a minute and use clips and text to tell simple but outlandish tales. Each day appears to have a different theme. On some days, celebrities accuse Democrats of leading their children to gender transition. Other waves of videos suggest Democratic candidates are engaged in embezzlement, child abuse or murder. Sometimes, they’ll hop onto breaking news – most recently, allegations that Democrats were behind the flydubai terrorist attack. The operation’s focus shifted to Europe for five days during the period that Bloomberg tracked.
Several posts pushed fabricated claims about Ossoff and Talarico, two of the most prominent Democrats in close races. One clip on September 11 pretending to be from CNN claimed Ossoff called for halting all military aid to Israel and redirecting resources to Ukraine. While Ossoff has voted to block specific weapons sales to Israel over concerns about civilian deaths in Gaza, he has supported continued US military aid.
Another video alleged Talarico would deny benefits to Hasidic parents who don’t allow their children to undergo gender transition. That claim is false. A single post that displayed both videos got 35,000 views, according to X’s count, despite being planted by an account with only 35 followers that previously had posted only in Japanese.
The stories about Talarico paying OnlyFans models and Ossoff owning undeclared properties illustrate the artificial view count spikes for Matryoshka videos. Uploaded on September 29, alongside another clip that falsely accused Alabama Democratic Senate candidate Everett Wess of committing sexual assault, the post got 50,839 views and only one retweet within 20 minutes of being published, according to snapshots archived by Antibot4Navalny and shared with Bloomberg. It got less than 700 views in the following three hours.
“Views don’t normally drop off a cliff,” Linvill said.
X, owned by Elon Musk, has taken steps to crack down on what it calls “co-ordinated inauthentic behaviour”. It suspended 80 per cent of the accounts seen by Bloomberg promoting Matryoshka content about the midterms. The platform removed 10 other accounts after Bloomberg flagged them. But even as accounts are suspended, new ones take their place. While Matryoshka also plants stories on other platforms such as BlueSky and TikTok, it most frequently posts on X, according to Antibot4Navalny.
“As much as X keeps taking them down, we always see a new crop,” said Jean le Roux, a senior investigator at Graphika, which tracks state-backed disinformation. “X is curbing the spread and the reach of these campaigns, but they are still there.”
X didn’t respond to emailed questions. It said in an October 14 post that it had suspended more than 13,000 accounts involving suspected foreign influence operations linked to Iran, Russia and China that targeted US political conversation.
While most Matryoshka posts have failed to gain real traction, a few have broken through. The Institute for Strategic Dialogue, a London-based research organisation that examined 135 pieces of content on social media between January and March 2025 that it attributed to Matryoshka, found one that garnered 4 million views: a video made to look like it was from E! News that falsely claimed the US Agency for International Development paid celebrities to travel to Ukraine. Musk and Donald Trump Jr each reposted the clip.
Most of the Matryoshka clips targeting the US midterms exhibit the hallmarks of co-ordinated activity by networks of bots. A September 24 video still on X alleges Minnesota Democratic senator Amy Klobuchar, who is running for governor, received $US72 million ($103 million) in suspicious donations – more than five times the amount she has actually raised, according to campaign filings. By X’s count, it got 155,300 views, two likes and no retweets, metrics that indicate the lack of a genuine audience in line with the view count.
Many Matryoshka clips try to portray the Democrats as extreme on transgender issues, a theme that piggybacks on Republican Party messaging in several midterm campaigns. Some promote a false narrative that Democrats are paying celebrities to promote transgender rights.
One, posted on September 22, falsely said The New York Times revealed that Taylor Swift accepted $US500 million to support the Democratic Party and gender transition in exchange for a series of concerts before the November election. It was published alongside a similar and also false story about singer Billie Eilish. The post got 75,600 views within 30 minutes of being uploaded and then quickly tailed off, according to snapshots archived by Antibot4Navalny.
“Even though they’re not actually making an impact on the US midterms or the way people vote, they’re probably still going to use this as a kind of a measure of success within Russia to report back to say how effective the campaign was,” Graphika’s le Roux said.
“They rely so much on those inflated metrics to prove their success.”
Bloomberg
Get a note directly from our foreign correspondents on what’s making headlines around the world. Sign up for our weekly What in the World newsletter.
