OpenAI has banned a cluster of ChatGPT accounts that very likely operated out of Russia[1]. Their job was to promote an "expert community" that claimed an address in Israel, while the site behind that name was largely stocked with other people's academic work. The reach was small, but the effort put into manufacturing credibility sets this case apart from the Russia-linked operations the company has disrupted before.

An Operation Built Around a Fake Expert Community

The banned accounts all pushed a brand called the International Burke Institute (IBI), which described itself as an expert community and listed a street address in Israel. The domain behind it was registered in February 2025[1].

The site was arranged to look like a roster of world-class researchers, and as of 17 July it named figures such as Francis Fukuyama and Noam Chomsky. Yet in a review of 36 articles published between September 2025 and May 2026, 34 turned out to be copied from elsewhere on the internet[1].

The copying was careless. An article on the China-Pakistan Economic Corridor appears to have been lifted from a Cambridge University Press original and reassigned to a professor whose field is South Asian politics. A piece on migration governance appears to have been taken from the Migration Policy Institute and credited to an Australian professor of food science[1]. Real content under fake bylines, in other words: enough to look the part while obscuring where the writing actually came from.

ChatGPT Handled Only the Promotion

What stands out is how narrow ChatGPT's role was. The articles on the site, and the reports built on the institute's own "sovereignty index," were not generated with OpenAI models[1].

The model was used for the social media posts that pointed back to those articles. Operators prompted in Russian, then published the resulting English-language comments on Substack, Telegram, X, Facebook and LinkedIn. Some prompts explicitly asked the model to strip out phrasing that would mark the authors as Russian speakers. Because OpenAI does not allow access from Russia, the operators connected through VPNs[1].

A second strand of activity produced German-language posts for a Telegram channel called "Lahme Ente" (lame duck), which criticized Ukraine, the EU and the German government while arguing for warmer relations with Russia. One operator also generated logos for about a dozen Telegram channels aimed at Germany, the United States, France, Poland and Türkiye, and repeatedly asked for Russian-language summaries of what those channels were posting[1].

Small Audience, Elaborate Scaffolding

By any measure of impact, the campaign underperformed. Individual posts drew low view counts and the official IBI accounts attracted few subscribers. The Telegram channels did better, generally in the range of 10,000 to 20,000 followers each[1].

Using the Brookings Institution's Breakout Scale, OpenAI places the operation at the lower end of Category Three: active across multiple platforms, with some signs of reaching authentic audiences[1].

There are also traces suggesting that a handful of real individuals in Israel represented IBI in article submissions and at conferences. OpenAI says it cannot determine how those people relate to the operators in Russia[1].

AI as Scaffolding, Not as the Main Act

The shape of this case argues against treating generative AI as the centre of modern influence work. Most of the text was human-assembled copy, and the model handled the peripheral task of advertising the storefront. What held the operation together was the three-part package placed in front of visitors: recognizable expert names, republished academic writing and a proprietary-sounding index[1].

The flip side is that this peripheral use is exactly what exposed the campaign. Generating promotional posts left a footprint on a platform that investigates such things, and the rest of the structure was traced from there. The carefully built foundation sat outside anything a model provider could flag on its own. That asymmetry marks the boundary of what enforcement by AI companies can and cannot reach.

Summary

OpenAI banned a cluster of Russia-based ChatGPT accounts that promoted IBI, a fake expert organization claiming to be based in Israel. Most articles on its site were copied from elsewhere, dressed up with well-known researchers' names and paired with a sovereignty index that flattered Russia. Actual reach stayed limited, but the construction of apparent authority was more elaborate than in earlier cases. AI handled only the promotional layer, and that layer is what gave the operation away.

[1] Source: https://openai.com/index/disrupting-malicious-uses-of-ai-influence-campaign-russia