California attorney general Rob Bonta has served an investigative subpoena on OpenAI, announced on 1 October and served the day before. His office says it is part of an ongoing California Department of Justice investigation into incidents resulting from the operations of OpenAI and its AI models. Bonta said last month that the department was conducting a formal investigation into the Hugging Face incident while continuing to monitor the industry’s compliance with state law more broadly.
Bonta put the question as one of responsibility rather than capability, saying frontier models can be legitimate tools for cyber defence, and that companies which develop them and offer them for use “have a moral and legal responsibility to ensure that they do not perpetrate or enable cyberattacks”.
Worth keeping straight: an investigative subpoena is a demand for documents and answers, not a charge. Nothing in the announcement alleges OpenAI broke any law, and the office has not said what it expects to find.
Why it matters: OpenAI published its own account of a coordinated model-distillation campaign two days earlier. A state regulator now asking the same company for its records narrows the distance between a lab disclosing on its own terms and having to account for itself on someone else’s.