OpenAI disclosed on Thursday that it had deactivated two influence operations that leveraged its ChatGPT platform to push disinformation, one originating from Russia and the other from Iran.
The Iranian operation operated seven “journalist” profiles that pretended to be American freelancers, producing roughly 100 long‑form pieces that appeared in about a dozen small to mid‑size news sites worldwide. The earliest article surfaced in July 2025; the latest ran in October 2026, with a spike in output after the U.S.–Iran war erupted earlier in the year.
The Russian effort ran under a front called the Social Research Center (SRC), a clandestine think‑tank that co‑opted Latin American staff who were unaware they were working for a Moscow‑based group. SRC’s content focused on undermining Ukraine’s reputation in the region and swaying local politics.
One of the Iranian personas, named Ervin B. Hoskins, maintained active X and Instagram accounts that the platform traced back to a West Asia Android app and a Tehran‑based base. OpenAI’s transparency data showed those accounts were suspended following the shutdown.
OpenAI said it will keep monitoring the platform for similar coordinated influence campaigns and will notify relevant authorities if new threats surface. The company’s action signals a tightening of its content‑moderation regime amid mounting concerns over state‑sponsored AI propaganda.