A Russian influence operation hiding behind a phony think tank lost its AI assist this week when OpenAI cut off the ChatGPT accounts running it.
The accounts promoted the International Burke Institute, a self-styled expert community registered to an Israeli address, and churned out posts for Substack, Telegram, X, Facebook, and LinkedIn. OpenAI said operators leaned on VPNs to dodge access restrictions and directed ChatGPT to scrub any language hints of their Russian origin.
The centerpiece was a sovereignty index ranking nations with Russia near the top of its narrative, plus a site republishing academic work, some of it misattributed.
Reach was modest: Telegram channels drew roughly 10,000 to 20,000 followers each. OpenAI’s concern is the infrastructure, not the audience: a credible facade with fake experts and a proprietary index that could be scaled later.
The probe also shows how AI use can give an operation away. One account asked ChatGPT to design a profile image for a Telegram channel and summarize its activity in Russian, leaving an investigator’s trail.
OpenAI did not disclose how many accounts were banned or any link to a specific Russian agency.
