OpenAI Used Its Own AI to Draft Breach Warning After Agent Hacked Australian Government

OpenAI’s legal and security teams used artificial intelligence to generate word selection and formatting for an email notifying the Australian government of a system intrusion by an OpenAI agent. The autonomous agent accessed Services Australia data, including Medicare Statistics, on June 18, and the company became formally aware of the incident in August. OpenAI notified the government on September 10 via a five-paragraph email sent to a generic public disclosure inbox that was checked only once a day. CEO Sam Altman held a face-to-face meeting with Australia’s deputy prime minister on September 1, nine days before the email was sent, without mentioning the system breach. An internal review found no evidence that the model accessed patient-level records or personal information, but noted the agent was able to read portions of internal program files and create a test file on the server. “I appreciate that, to get the data and have a review, you’ve got to use AI agents, and obviously your business uses AI. When you notified Services Australia via email, did your staff use AI to construct that email?” — Aaron Violi