OpenAI Bans Russian Accounts That Used ChatGPT to Build a Fake Academic Institution

Key points

• OpenAI banned a cluster of ChatGPT accounts in 2025 linked to a covert pro-Russia influence operation that spread content across at least five social platforms.
• Operators used VPNs, software that masks a user's real location by routing their internet connection through another country, to bypass OpenAI's ban on Russian access to its models.
• The campaign created a fake organisation called the International Burke Institute, complete with copied academic articles and false author credits.
• OpenAI said the operation's reach was limited, but warned that the infrastructure it built could be scaled up over time.
• This is the second time this year OpenAI has shut down Russian-linked ChatGPT accounts used for influence operations.

OpenAI has banned a group of ChatGPT accounts it traced to Russia, saying they were being used to run a covert campaign designed to spread misinformation and make pro-Russia content appear to come from legitimate academic sources.

The company says it spotted the operation after investigating AI-generated posts on social media. That inquiry, as OpenAI described in a blog post this week, led to what it called a "much broader influence operation."

What was the operation actually doing?

The operators were writing social media comments in Russian, feeding them to ChatGPT, and posting the outputs on Substack, Telegram, X, Facebook and LinkedIn. They specifically instructed the chatbot to strip out any phrasing that might reveal the text was written by a Russian speaker.

The content pushed readers toward an organisation called the International Burke Institute, or IBI. Its website looked professional. It carried articles that appeared to be peer-reviewed academic work. Many of those articles were copied from real researchers, sometimes with the original author's name swapped out for a fake one.

The campaign also built what it called a "sovereignty index," a ranking that praised Russia and criticised Western countries. The index was presented as objective analysis.

OpenAI said the operation's actual audience was small. But the company's concern is less about this campaign's reach and more about what it had constructed: a believable institution with invented experts, borrowed credibility and a custom data product, all ready to be amplified.

"This illustrates how influence actors can use AI as a supporting tool within a broader effort to manufacture authority," OpenAI said.

Should ordinary readers be worried?

Not about this specific operation, which appears to have had little real-world impact. The bigger takeaway is the template it reveals.

A think-tank with a professional website, published research and a proprietary index looks credible. The effort required to build that kind of front used to be significant. AI now makes it far cheaper and faster.

First reported by CNBC Tech, the story also connects to a February incident in which OpenAI shut down ChatGPT accounts linked to Rybar, a pro-Russia media company the UK government describes as partially coordinated by the Russian Presidential Administration and funded by state corporation Rostec.

OpenAI does not permit access to its models inside Russia. This group got around that restriction using VPNs.

What should you watch for?

Before trusting a think-tank, index or research institute you have not heard of before, check a few things. Search the named experts independently. Look for the original source of any research they cite. Check whether the organisation appears in any independent news coverage. A polished website and a list of publications are not, on their own, evidence that an organisation is real.

If a social media post points you to an unfamiliar institute that happens to rank countries or rate geopolitical risk, treat it with the same scepticism you would apply to an anonymous source.