Deepfake Celebrities and the Midterms: How the Russian Campaign Works and How to Spot It

Top congressional Democrats on intelligence oversight say a classified US intelligence report shows Russia running an AI-assisted online campaign to discredit candidates before the November 3, 2026 midterms. Researchers have tied deepfake celebrity videos and fake "news" clips on X to two Russia-linked networks, Matryoshka and Storm-1516. Officials report no evidence of hacked election systems, and no Russian response to the latest claims has been reported.

What did US lawmakers say about Russia and the midterms?

They said a classified US intelligence report shows Russia is using AI tools to spread disinformation about individual candidates and erode trust in the election. Sen. Mark Warner and Rep. Jim Himes, the senior Democrats on the Senate and House intelligence committees, described the report to Politico on October 21. According to Reuters, the briefing they requested came from Director of National Intelligence Jay Clayton.

• Who may be targeted: the lawmakers believe Russia is going after Democrats and Republicans who support Ukraine.
• Named candidates: Reuters reported Warner citing news reports that Talarico and Georgia Sen. Jon Ossoff are being hit by AI-generated fake ads. Cybernews, also citing Politico, wrote that the lawmakers did not name specific targets, so accounts of that detail differ.
• What was not found: Himes stressed there is no evidence any country can penetrate US election systems or change results.
• Tone: Himes called it "the same old, same old," per Reuters.

The White House, ODNI and the Russian embassy did not immediately respond to Reuters. Moscow has denied meddling in past US elections.

What were the deepfake celebrity videos?

They were AI-fabricated clips that made Robert De Niro, Julia Roberts and Cher appear to attack Democrats. Some urged viewers to join a #AllDemocratsAreCriminals campaign. According to Bloomberg, the clips began appearing on X in early September, and researchers linked them to a Russian state-backed operation known as Matryoshka, also called Operation Overload.

The operation also imitated journalism. Bot-tracking researchers cited by Meduza said the clips were posted under the logos of real news outlets. Kyiv Post's summary of the Bloomberg investigation says Bloomberg reviewed 130 fabricated videos and doctored newspaper front pages circulated between September 10 and October 9. They included clips styled with ABC News and CBS News branding that made unsupported claims about Talarico and Ossoff. The same summary reports the network targeted Democratic candidates in 20 of 35 Senate races.

A related case shows how brand impersonation works in practice. The AI Incident Database logs a Matryoshka deepfake falsely presented as a Mother Jones production that impersonated its editor. X removed the posts after Mother Jones complained.

How were the view counts faked?

The numbers were pumped up so the posts looked far more popular than they were. Bloomberg reported that some clips reached about 200,000 views in their first hour and then stopped growing, while drawing almost no likes or reposts. Forensic data from the researcher group Antibot4Navalny, cited in the Kyiv Post summary, showed that pattern of a sudden spike followed by a plateau. Real engagement usually builds more gradually. The amplifying accounts had often been dormant for years and had few followers.

This matters because a big view count signals social proof. Seeing "200K views" can lead people to assume a clip is widely believed, even when almost nobody actually engaged with it.

What is Storm-1516 and the Talarico video?

Storm-1516 is a separate Russia-linked network that researchers say was behind videos posted on October 20 that baselessly accused Talarico, the Texas Democratic Senate nominee, of misconduct. NBC News reported that researchers at Graphika, Alethea and Clemson University linked the videos to Storm-1516, a group researchers have tied to Russian military intelligence.

Reported reach varies. Meduza, citing NBC, put one post above 1.3 million views. The New York Times, as cited by the San Antonio Current, reported more than 850,000 views in the first 24 hours. A Talarico campaign spokesman told the Times that opponents are relying on AI deepfakes. Researchers say the network made similar unfounded claims about Minnesota Gov. Tim Walz in 2024.

How have X and critics responded?

X says it is removing foreign influence networks, and Republicans have criticized how the intelligence was disclosed. On October 15, X said it had suspended more than 13,000 accounts tied to three suspected influence operations linked to Iran, Russia and China. X said those accounts targeted candidates "across the political spectrum." No X statement about the specific celebrity deepfakes was found in the coverage reviewed for this post.

House Intelligence Committee Chair Rick Crawford accused the Democrats of "cherry-picking" classified details for the press, according to Cybernews. He also said they highlight Russia while ignoring China. Himes separately told Reuters he had no reason to believe China was not running targeted efforts too.

How can you spot a deepfake before you share it?

The most reliable defense is checking the source, not studying pixels. Media-literacy groups such as the Freedom Forum recommend the SIFT habit: Stop, Investigate the source, Find better coverage, Trace the claim to its origin.

1. Pause on strong emotion. Content built to make you angry about a candidate is the kind most worth checking.
2. Check who posted it. A new or long-dormant account with few followers posting a "news" clip is a red flag.
3. Verify the logo. If a clip carries a major outlet's branding, look for it on that outlet's own site or official account.
4. Look for other coverage. A real celebrity endorsement or a real scandal will be reported by several credible outlets.
5. Compare views with engagement. A huge view count with almost no likes, replies or reposts fits the inflation pattern Bloomberg described.
6. Notice what feels off. Odd lip-sync, flat voice tone or unnatural blinking can be clues, but good fakes may show none of them.
7. When unsure, don't share. Check fact-checkers and researcher reports first.

What should brands learn from fake videos under real news logos?

Any recognizable logo or spokesperson can be borrowed to make a lie look credible. Matryoshka did not need to hack ABC, CBS or Mother Jones. It only had to copy their look. That is the brand-safety lesson for every organization with a trusted name.

Practical steps for marketing and communications teams:

• Monitor for impersonation, not just mentions. Track your logo, visual style and executives' faces and voices across short-video platforms, not only your brand name in text.
• Vet the voices you pay. Clemson researcher Darren Linvill told Straight Arrow News the Talarico post was pushed by marketing bots and then paid influencers. A creator audit that checks partners' audience authenticity and engagement quality reduces the chance your brand ends up next to inflated or coordinated activity.
• Know how AI systems describe you. When fake clips spread, they can shape how AI assistants summarize a brand or person. Tools like Model Share and an AI brand intelligence practice help you see what models say about you. Our guide to auditing how AI describes your brand covers the steps.
• Prepare a takedown path. Mother Jones got posts removed by complaining to X. Know your platform reporting channels and keep official spokesperson accounts verified before you need them.

FAQ

Is Russia confirmed to be behind the deepfake celebrity videos?

Independent researchers attribute the celebrity clips to the Russia-linked Matryoshka network, and Warner and Himes say a classified US report shows a Russian AI-enabled campaign. The underlying intelligence has not been made public, and Russia has denied meddling in past US elections.

Have Russian hackers changed any election systems?

No. Lawmakers say there are no proven attempts to hack election systems or campaign networks, and Himes said there is no evidence any country can change results.

Were the claims in the Talarico videos true?

Researchers describe the accusations as baseless, and his campaign called the attacks AI deepfakes. NBC News reported that researchers linked the videos to the Storm-1516 network.

Why did the fake videos have so many views?

Bloomberg reported the view counts were artificially inflated by bots. Some clips hit about 200,000 views in the first hour and then plateaued, with almost no likes or reposts.

What has X done about it?

On October 15, X said it suspended more than 13,000 accounts tied to suspected influence operations linked to Iran, Russia and China. It also removed a Matryoshka deepfake impersonating Mother Jones after the outlet complained.

What is the simplest way to check a suspicious political video?

Look for the same story from the outlet whose logo appears on the clip, and from other credible news sources. If no one else has it, don't share it.

Last updated October 24, 2026.