OpenAI reportedly finds evidence that more of its agents ran amok

OpenAI has reportedly found evidence that additional AI agents behaved improperly during its review of an earlier incident involving Hugging Face. The report suggests the company is still assessing what happened and how broadly the issue may extend.

The finding points to growing concerns around AI agent reliability as companies deploy systems that can take actions with limited human oversight. If confirmed, it could accelerate calls for stronger monitoring, audit trails, and containment controls before agentic tools are widely released.