OpenAI published a report on August 25, 2026, describing the disruption of a covert influence operation that used ChatGPT-generated posts to promote the International Burke Institute (IBI). The company said it banned a cluster of ChatGPT accounts that very likely originated in Russia and that operators used VPNs to access the platform because OpenAI does not allow access from Russia.
Structure and tactics of the operation
According to OpenAI, operators prompted the models in Russian to produce social media comments in English and instructed the system to remove linguistic clues of Russian origin. Generated posts appeared on X, LinkedIn, Facebook, Substack and Telegram, sometimes from accounts using the IBI name and logo and sometimes from likely inauthentic everyday-user accounts. Operators also used ChatGPT to draft replies on Substack asking readers to follow the IBI channel.
The IBI website, registered in February 2025 and presented as an Israel-based “expert community,” published reports and an ostensible “sovereignty index” that portrayed Russia positively and criticized Western countries. OpenAI reported that many site articles were copied or misattributed: in a sample of 36 articles published between September 2025 and May 2026, 34 were taken from other sources. One example cites an IBI item dated December 28, 2025, attributed to Professor Katharine Adeney but apparently copied from a Cambridge University Press piece by Yunas Samad published July 7, 2025. Another item copied a Migration Policy Institute report from April 2025 and was misattributed on the IBI site.
Scale and consequences
OpenAI judged the campaign’s immediate audience reach to be limited: official IBI accounts had low subscriber numbers and typical posts received few views. The report noted, however, that several Telegram channels linked to the operation held roughly 10,000–20,000 followers each. Using the Brookings Breakout Scale, OpenAI assessed the operation at the lower end of Category Three, meaning activity occurred on multiple platforms with some signs of breakout to authentic audiences.
OpenAI highlighted the operation’s broader infrastructure — a credible-appearing website, republished academic work, a manufactured index and social accounts — as the context in which AI-generated promotional posts were used. Concrete actions reported by OpenAI include banning the implicated ChatGPT accounts; the company also identified examples of non-idiomatic English and machine-translated phrasing on the IBI site and related channels such as the Telegram channel “Lahme Ente.”
Original source: OpenAI News