OpenAI bans a fake Russian think tank
Archive item — written before sources were shown.
OpenAI banned Russia-linked ChatGPT accounts that used a fake Israel-based think tank to push a pro-Russia 'sovereignty index' across social platforms.
OpenAI disclosed August 25 that it banned a cluster of Russia-linked ChatGPT accounts behind a previously unreported influence campaign. The operators worked in Russian and routed connections through VPNs to get around OpenAI’s access restrictions for Russia, then used ChatGPT to draft social media content that ran on Substack, Telegram, X, Facebook, and LinkedIn.
The centerpiece of the campaign was a fake organization called the International Burke Institute, a self-described expert community whose website listed an Israeli street address and falsely claimed academics including Francis Fukuyama and Noam Chomsky among its ranks. OpenAI’s review of 36 articles tied to the site’s supposed experts, published between September 2025 and May 2026, found 34 were copied from elsewhere online, sometimes with the wrong attribution. The operation’s main product was a “sovereignty index,” a metric built specifically to rate Russia favorably and criticize Western nations, spread through the fabricated think tank’s supposed credibility.
What it means for you
This is one more entry in a pattern operators should already be tracking: state-linked influence operations increasingly use commercial AI tools for the unglamorous work, drafting, translating, and distributing content at scale, then hide behind a fabricated institution to launder credibility. If your work touches trust and safety, content moderation, or anything that ingests third-party expert commentary, the actionable lesson is the same one OpenAI’s own disclosure demonstrates: a fake author bio and a professional-looking website are cheap to produce and easy to check, so verify institutional identity independently before treating any “expert” source as credible, especially one you found through social media rather than direct outreach. It also reinforces why detecting AI-generated text alone is the wrong signal to gate on; the tell here wasn’t the writing style, it was the copied articles and the fabricated affiliations underneath it.
- 01Disrupting a new covert influence campaign from Russiaopenai.com · primary
- 02OpenAI bans Russian ChatGPT accounts used in covert misinformation campaigncnbc.com · reporting
