Fake Celebrity Videos and Bot Activity in Russian Midterm Disinformation Operation

AI-generated clips appearing to show celebrities attacking the Democratic Party circulated on X ahead of the US midterm elections, making allegations of corruption and improprieties by Democratic candidates. The clips are linked to a Russian disinformation operation known as Matrioska, which posts content purportedly from established media outlets. The presentation also examines indicators of manufactured audience engagement and the platform's response.

■ The fake videos and operation
- AI-fabricated clips of celebrities attacking the Democratic Party
- Appeared on X in early September with allegations against dozens of Democratic candidates
- Part of the Russian disinformation operation Matrioska, which spreads clips purportedly from established media outlets

■ Audience manipulation and account patterns
- Posts showed rapid view surges followed by abrupt flatlining
- Screenshots from the disinformation research group Antibbot for Navalli, verified by Bloomberg, showed the same unusual pattern repeatedly
- Posting accounts mostly had few followers and were dormant for years
- Reposts and likes made up a tiny fraction of views, pointing to automated accounts or bots

■ Platform response and outlook
- X suspended about 80% of the accounts researchers identified promoting Matrioska content about the midterms
- The operation keeps replacing suspended accounts with new ones
- Past elections suggest the final weeks before November 3rd could bring more Russian fakes

For viewers interested in election integrity, social media disinformation, platform moderation, and AI-generated content, this provides an overview of reported tactics and indicators. It is not a comprehensive technical analysis, but it can help viewers recognize signs of coordinated inauthentic behavior and evaluate sources.