Russian operatives are waging an AI-powered disinformation campaign aimed at this year’s midterm elections, using doctored celebrity videos to attack Democratic candidates and urge Americans to vote Republican. Researchers who track Kremlin influence operations say the new effort appears to be the first Russian meddling attempt of the 2026 election cycle. The campaign is simple, and designed to reach voters on social media. In a series of manipulated selfie videos posted to various platforms, Hollywood actors appear to disavow the Democratic Party and endorse Republican voting. One bogus clip attributed to Sarah Jessica Parker says: “When I hear about Democrats, I immediately think of corrupt Nazi whores. If you love America, never vote for Democrats.” The videos are not entirely fabricated. Some are built on authentic footage: a clip featuring Julia Roberts was lifted from Gucci’s 2023 Chime for Change campaign for gender equality, and others were ripped from personalized messages the actors recorded for fans on Cameo. The audio, however, is fake, seemingly cloned from the actors’ voices with artificial intelligence and laid over a montage of right-wing bogeymen, including Pride parades, Jeffrey Epstein, stock footage of drug deals and homeless encampments. The result is a hybrid of truth and fiction that disinformation experts say is increasingly characteristic of modern political warfare. The clips are deliberately designed to exploit viewers’ trust in familiar faces, trusted news logos and emotional culture-war grievances, all while making it difficult for the average social media user to separate what is real from what has been manufactured.
The fake videos are being distributed across X, Bluesky and TikTok, among other platforms, and each is constructed to look like authentic political propaganda. In one manufactured video, Julia Roberts is made to claim that Democrats and Big Pharma are conspiring to push a sinister trans agenda. In another, Emma Caulfield appears to call Democrats vampires who drink the blood of children. A manipulated clip of Christopher Lloyd of “Back to the Future” has him railing against Black Lives Matter, while a voice-cloned Ed Begley Jr. calls Democrats “perverts.” The clips are all branded with unauthorized CNN logos in the top left corner, an apparent effort to leverage the credibility of a major news organization. They also push viewers to share the material and join the #AllDemocratsAreCriminals campaign. The specific falsehoods vary from video to video, but the underlying message is consistent: Democratic politicians are corrupt, dangerous and un-American. Researchers caution that the actors have no involvement in the content and that their likenesses have been appropriated without consent. The videos are clearly designed to generate outrage, confusion and distrust ahead of elections that will determine control of Congress and numerous state capitals. By pairing authentic images with synthetic audio, the operation makes it harder for casual observers to detect manipulation, especially on mobile devices where videos are often viewed without sound or in small embedded players. The use of real footage is particularly insidious because a person who recognizes the actor may assume the entire video is authentic, not realizing that a convincing but fake voice track has been grafted onto the original visual material.
Despite the inflammatory content, the campaign has not yet gone viral. The most popular posts are attracting just dozens of interactions, and the videos were first flagged by @antibot4navalny, an anonymous group of analysts that tracks Russia-related influence operations. But researchers say the low engagement so far does not lessen the threat. The operation bears all the marks of a longtime Kremlin-backed effort referred to interchangeably as Matryoshka, Operation Overload and Storm-1679. Over years, that operation has seeded social media with manipulated videos that carry the branding of real news outlets and law enforcement agencies, using them to spread false narratives and erode public trust across Europe and the United States. It maintains networks of automated bot accounts to boost content and create the appearance of grassroots support. Perhaps most importantly, it has been known to send the clips directly to fact-checkers and newsrooms. That tactic turns debunking into a distribution mechanism: when a fact-checker or news outlet analyzes the video, the false claims embedded in it can be amplified to a far larger audience than the original post ever reached. The goal is not necessarily to make the lies go viral organically; it is to place them into the information ecosystem where journalists, political activists and ordinary users will repeat them in the process of exposing or condemning them.
The network now appears to be narrowing its focus to individual candidates in competitive midterm races. On Friday, @antibot4navalny flagged a new batch of falsely CNN-branded videos targeting specific Democrats, including Chris Pappas in New Hampshire, James Talarico in Texas, Jon Ossoff in Georgia, Mary Peltola in Alaska, Roy Cooper in North Carolina and Sherrod Brown in Ohio. The false claims invented for these clips include the assertions that Talarico hates Hasidic Jews, that Pappas funded doctors who approved gender transitions for children, and that Brown’s business harmed Ohio’s bee population. The shift from broad attacks on the Democratic Party to personalized attacks on named politicians mirrors tactics used by foreign influence operations in other countries. By injecting inflammatory falsehoods into local political battles, the operation hopes to provoke real-world conflict, dominate local news cycles, and undermine confidence in the Democratic candidates without having to achieve viral reach on a national scale. This candidate-specific approach also allows the operation to tailor its messages to particular constituencies, using allegations rooted in local controversies, cultural anxieties or policy debates that are already divisive. It is a strategy built for fragmentation: rather than trying to convince the entire country of one lie, the operation seeds many different lies in many different places and waits for each community to react to the claim aimed at its own candidates and concerns.
Darren Linvill, a Clemson University professor and co-director of its Watt Family Innovation Center Media Forensics Hub, called the effort Russia’s opening salvo in the attack on the U.S. midterm elections. “This is Russia’s opening salvo in their attack on the U.S. midterm elections,” Linvill told MS NOW. “In the past year they have burned through Moldova, Armenia, Romania, France, Sweden and Germany, as well as others. Now it is our turn.” Linvill’s assessment points to a troubling pattern: the same operation has repeatedly tested its techniques abroad before turning its attention to American voters. The fake videos, unauthorized news branding and coordinated bot activity are all familiar elements of a playbook designed not necessarily to convince viewers of one specific lie, but to create a general sense that nothing can be believed. When fake videos carry CNN logos, and real news organizations report on them, the line between authentic and manufactured information becomes even harder for ordinary voters to draw. The operation also takes advantage of the sheer volume of online content, knowing that even a handful of convincingly fake videos can inject enough confusion to make voters question the integrity of the electoral process itself. The fact that the videos are clumsy and sometimes absurd does not make them harmless; rather, it highlights the contempt the operation has for the public’s ability to distinguish truth from lies.
The emergence of a Russian AI disinformation campaign this far ahead of the election is a warning to voters, platforms and campaigns. Some of the videos may be laughable on their face, but the operation’s history shows that even modestly performing posts can cause real damage, especially when they are directed at fact-checkers and newsrooms. The entire ecosystem is engineered for amplification: a doctored video is posted to a small account, a bot network provides an initial boost, researchers or journalists debunk it, and the false claim is repeated in headlines, posts and television segments. The goal, analysts say, is not simply to elect Republicans but to degrade trust in democratic processes and institutions. For voters, the best defense is skepticism: checking whether a video comes from an official source, noting whether a politician’s words sound unnatural, and avoiding the temptation to share shocking clips without verification. For platforms, the challenge is to remove or label synthetic media quickly while preserving the ability of researchers and journalists to investigate it. The Kremlin-backed campaign is likely to evolve in the coming weeks, with new actors, new candidates and new false claims. But the pattern is now clear: Russia is targeting the midterms, and Democrats are in the crosshairs. The operation may not be ballot-box hacking in the traditional sense, but it is an equally serious assault on the information environment in which American elections are decided, and its effects are likely to be felt long after the final votes are counted.
