An AI agent exported a patient record. Your logs can't say who told it to. (opens in new tab)

You put an LLM agent into production. It runs under a service account or a shared API key, because that's how you give software credentials. It reads a record, exports a file, moves some money. Your audit log dutifully records the action. It says the agent did it. It does not say which human told it to. That's fine right up until it isn't. If the agent does something it shouldn't have, "the service account did it" is not an answer anyone can act on. You can't discipline a service account. You...