California gives OpenAI a new prompt: show your work
California served OpenAI an investigative subpoena on September 30. The inquiry is real; whether it establishes legal wrongdoing remains an open question.
The next prompt comes with a letterhead, droogies.
The new paperwork
California Attorney General Rob Bonta served OpenAI with an investigative subpoena on September 30, according to the California Department of Justice’s October 1 announcement. It concerns cybersecurity incidents and risks involving the company and its AI models.[1]
This is a new step in an ongoing inquiry. The department says it had already announced a formal investigation of the Hugging Face incident last month.[1]
The unresolved bit
Bonta says his office is determining whether developers should be held legally accountable in this case. The announcement reports an investigation and a subpoena. It announces no finding of wrongdoing, penalty or judgment.[1]
Reuters corroborated the subpoena on October 1 and said OpenAI had not immediately responded to its request for comment.[2]
The droog take
A splendidly unglamorous milestone for the age of autonomous intelligence: somebody wants the paperwork.
The useful questions are concrete. What authority did the agents have? Which controls constrained them? Where, if anywhere, did the controls fail? Those are questions for the evidence, however much less exciting they sound than a machine with a mind of its own.
Viddy what the inquiry establishes, what the company supplies and what changes afterward. The subpoena gives this story a new document. It doesn’t write the ending.
FOLLOW THE SOURCES.
Checked Oct 2, 2026. Numbered references point to the evidence behind factual passages. Commentary is our own.
Microsoft wants autonomous agents working on your desktop. It also shipped an OS-enforced container to stop them wandering through the files. Progress has acquired a lock on the medicine cabinet.