Overview
- OpenAI published a threat report and banned two clusters of ChatGPT accounts on Wednesday, June 10, after finding they were used to generate social posts, comments and political cartoons tied to covert influence efforts.
- Researchers labeled the clusters “Data Center Bandwagon” and “Tech and Tariffs,” with one traced to a Chinese technology firm that does government work and the other linked to actors OpenAI says likely originated in China.
- Both operations prompted ChatGPT in Simplified Chinese, used VPN access to reach the service, asked for English and other language outputs, and posted material through likely inauthentic accounts posing as Americans on platforms like X and YouTube.
- OpenAI found little authentic engagement and no evidence the campaigns shifted public opinion, while also refuting false claims circulated by the networks that ChatGPT user data had been compromised.
- The episode highlights how generative models can cheaply scale influence testing, deepens U.S.-China tech tensions, and has already drawn calls from lawmakers for investigations and possible legislation on AI and data‑center policy.