OpenAI bans ChatGPT accounts behind Russian influence campaign pushing a fake think tank

OpenAI said it has banned a cluster of ChatGPT accounts tied to a Russia-origin influence operation that used the chatbot to help promote a fabricated think tank and spread anti-Ukraine, anti-EU messaging across multiple social platforms.

What's new

In a threat-intelligence report published August 25, OpenAI wrote: "We recently banned a cluster of ChatGPT accounts originating in Russia that were being used to promote the International Burke Institute (IBI)."

According to OpenAI's writeup, the operators used ChatGPT to:

• Draft articles and other written material for the IBI website, a think tank OpenAI describes as fake, including content that appears to reuse or plagiarize existing academic writing.
• Generate social media posts promoting the IBI site, distributed across X, LinkedIn, Facebook, Substack, and Telegram.
• Produce German-language posts criticizing Ukraine and the European Union.
• Create logos for a set of affiliated Telegram channels.

OpenAI's response was to ban the ChatGPT accounts identified as part of the operation.

Context

This is the latest in a recurring series of reports OpenAI publishes on state-linked and other coordinated actors caught using its models for influence operations, scams, or malware development. Earlier reports in the series have covered network takedowns tied to Russia, China, Iran, and North Korea, typically describing the tradecraft used — AI-drafted articles, translated propaganda, fake personas and logos — alongside the accounts OpenAI removed.

The International Burke Institute itself appears to be an invented entity, built specifically to launder narratives into the appearance of independent policy analysis. Using a chatbot to mass-produce plausible-sounding academic and social content is a pattern OpenAI and other labs have flagged repeatedly as a low-cost way to scale propaganda operations that once required larger human staffs.

Why it matters

The takedown is a small, self-reported action rather than a major security incident, but it is a useful data point for how frontier AI labs are handling the political-abuse side of safety enforcement. Generative tools lower the cost of producing convincing, multilingual propaganda and fabricated institutional cover stories, which is exactly the capability OpenAI describes being used here.

For policymakers and platform trust-and-safety teams, these disclosures are also one of the few public windows into how AI companies detect and respond to state-linked misuse of their own products — evidence that will matter as governments weigh new disclosure or auditing requirements for frontier model providers.