OpenAI Uncovered A Russia-Linked Influence Campaign. Operators Used ChatGPT To Look Less Russian.

OpenAI has banned a cluster of Russia-linked ChatGPT accounts that were used in a covert influence campaign built around a purported Israeli think tank, copied academic work and social media posts designed to conceal the operators' Russian origins.

The accounts very likely originated in Russia and used VPNs because OpenAI does not make its models available there, the company said in a report published Tuesday. Operators typically prompted ChatGPT in Russian but generated mostly English-language comments for Substack, Telegram, X, Facebook and LinkedIn, at times explicitly asking the chatbot to remove linguistic clues that could reveal where they were based.

The campaign centered on the International Burke Institute, or IBI, which described itself as an Israel-based "expert community." Its website, registered in February 2025, featured prominent academics and intellectuals and published research that gave the operation the appearance of an established institution.

OpenAI's review found that appearance was largely manufactured. Of 36 articles published between September 2025 and May 2026 that it examined, 34 had been copied from elsewhere online. Some were years old and others were attributed to the wrong authors. One article originally published by Cambridge University Press was assigned to a different academic, while another piece from the Migration Policy Institute was attributed to an Australian food-science professor.

Le Monde reported that the IBI also claimed links to organizations including the United Nations and UNICEF.

UNICEF told the newspaper that the institute was not an approved partner and that it had no record of working with it. Le Monde also found that several people presented as connected to the institute lived in Israel and had Russian backgrounds, though OpenAI said it could not determine the relationship between individuals in Israel, the IBI and the operators in Russia.

A major feature of the site was a "sovereignty index" that portrayed Russia favorably while criticizing Western countries. The institute published reports targeting France, Germany, the United States and the European Union, while ChatGPT-generated social posts were used to direct readers toward the material.

The network stretched beyond the IBI. One operator used ChatGPT to create German-language posts for a Telegram channel that criticized Ukraine, the EU and Germany while advocating closer relations with Russia. Another generated logos for about a dozen Telegram channels focused on countries including the U.S., France, Germany, Poland and Türkiye, according to OpenAI.

OpenAI said the campaign appeared to have struggled to build a large audience, with typical social media posts receiving few views and official IBI accounts attracting relatively small subscriber numbers, although associated Telegram channels generally had between 10,000 and 20,000 followers.

The takedown follows earlier cases involving Russia-linked actors using generative AI for influence operations. In February, OpenAI banned accounts associated with the pro-Kremlin Rybar network after finding ChatGPT had effectively served as a content farm producing posts and comments in several languages. Ukraine's Center for Countering Disinformation said the network also used AI to develop proposals for influence campaigns in Africa.

OpenAI has previously found that covert influence operators generally combine AI with traditional tools such as websites and social media accounts rather than relying on AI alone. In an earlier threat assessment, the company said Russia-origin operations it disrupted had used its models for tasks ranging from generating political comments and articles to translation, proofreading and basic coding.