
OpenAI banned Russia-origin accounts using AI to promote a fake Israel-based think tank and a “sovereignty” index praising Russia and criticizing the West. Much of the generated content was to promote a self-described “expert community” called the International Burke Institute (IBI). Screenshot from an article about Germany on the IBI website. The operation’s main use of ChatGPT was to generate social media posts that promoted articles on the IBI website. As well as generating content about IBI, one of the operators generated German-language posts that were posted on a Telegram channel called “Lahme Ente” (“lame duck”).
Our mission is to ensure that artificial general intelligence benefits all of humanity. We advance this mission by deploying our innovations to build AI tools that help people solve hard problems. This includes building tools that enable us to detect, investigate, disrupt and expose covert influence operations (IO): deceptive attempts to manipulate public opinion or influence political outcomes without revealing the true identity or intentions of the actors behind them.
We recently banned a cluster of ChatGPT accounts originating in Russia that were being used to promote the International Burke Institute (IBI), a self-described “expert community” based in Israel. What began as an investigation into AI-generated social media posts led us to a much broader influence operation, built around a website containing copied and misattributed academic work, a “sovereignty” index called the Burke Sovereignty Index that cast Russia in a favorable light, and efforts to disguise the operators’ Russian origins. Although the campaign appears to have reached relatively small audiences, its elaborate construction distinguishes it from other Russia-linked (opens in a new window) influence (opens in a new window) operations (opens in a new window) we have disrupted (opens in a new window) since the start of the war in Ukraine. This report explains how the operation worked, where ChatGPT fit into it, and what we did to disrupt it.
We banned a cluster of ChatGPT accounts that very likely originated in Russia. The operators prompted in Russian to generate social media comments that were posted on Substack, Telegram, X, Facebook and LinkedIn. Most of the comments they generated were in English, and the operators instructed ChatGPT to hide any linguistic clues that they were Russian. As we do not allow access to our models from Russia, they used VPNs to access our platform.
Much of the generated content was to promote a self-described “expert community” called the International Burke Institute (IBI). The IBI brand was associated with a website which was registered in February 2025. This site claimed that the Institute was based in Israel. Articles published on the site were not generated from our models. Many were copied from real academic writings, sometimes with false attribution; some appear to have been drafted by a Slavic speaker and machine translated. As an example, the screenshot below from the IBI website uses an unnatural phrase that is most readily explained as a literal machine translation from a Slavic language.
Screenshot from an article about Germany on the IBI website. The phrase “the Svetofor coalition” is meant to refer to Germany’s coalition of socialists, liberals and greens, known from their party colors of red, yellow and green as the “traffic-light coalition” or, in German, “Ampelkoalition”. The word “svetofor” (светофор) means “traffic light” in a number of Slavic languages, including Russian, but is exceptionally unlikely to occur to an English or German speaker to describe the coalition: it was likely machine translated from a Slavic original.
This is, to the best of our knowledge, a new and previously unreported covert influence campaign. Its most notable feature is its creation of the “sovereignty index” promoted by the ostensible “think tank” to praise Russia and denigrate Western countries. This is the first time we have disrupted an influence operation tied to Russia that went to such elaborate lengths.
Using the Brookings Breakout Scale (opens in a new window) to assess the impact of this operation, we would assess it as at the lower end of Category Three (multiple platforms, with some indications of breakout to authentic audiences).