California Attorney General Rob Bonta served OpenAI with an investigative subpoena on September 30. His office announced the step on October 1; Reuters also reported it.
The department is asking additional questions about cybersecurity incidents and risks involving the models. This continues an existing investigation that includes the Hugging Face incident.
Bonta argues that developers bear safety responsibilities during training and testing as well as deployment. Whether OpenAI is legally responsible remains a matter for investigation.
This risk signal records an official measure, not a judgment or another autonomous AI incident. The reviewed announcement does not establish the full questions or an OpenAI response.