OpenAI has detected and banned a cluster of suspected China-linked accounts that were using ChatGPT to run a covert influence operation targeting Americans, according to reporting by TechSpot and TechRadar.
The now-banned accounts were not hacking infrastructure or stealing secrets — they were generating persuasive content designed to shape public opinion. According to TechSpot, OpenAI wrote that the accounts were "supporting covert influence operations that promoted narratives in an attempt to manipulate a legitimate debate about American AI" and data center policy.
The specific target was the ongoing national conversation about where and how the US should build out AI data centers — a debate with real stakes involving land use, energy consumption, local community impact, and national competitiveness. By seeding narratives critical of US data center expansion, the operation appeared aimed at slowing or complicating American AI infrastructure development.
OpenAI did not publicly detail the exact content produced or the scale of the operation, but the disclosure follows a pattern the company has established of publishing threat intelligence reports on how its tools are misused by state-linked actors.
The case is a clear example of AI being turned against the interests of the country that built it — and a reminder that influence operations no longer require armies of human writers. Cheap, scalable text generation lowers the barrier for foreign actors to flood domestic debates with manufactured opinion, making detection and transparency from AI companies increasingly critical.