A report published October 8 says AI helped draft parts of OpenAI’s September 10 notification to Services Australia, citing a person familiar with the incident. The person said OpenAI’s legal and security teams used AI for wording, word choice and formatting, while people reviewed and sent the final email.
Jason Kwon’s answer to Parliament
At a parliamentary hearing on October 6, Aaron Violi asked OpenAI chief strategy officer Jason Kwon whether staff had used AI to construct the email. Kwon said he did not believe they had, and that OpenAI would confirm. He also acknowledged that the company’s response was inadequate and that it should have notified affected parties sooner.
Kwon discussed the notification process during the hearing, explaining that staff had treated the incident as a technical matter and sought to contact technical counterparts.
What OpenAI’s notice said about the model’s activity
OpenAI’s notification described activity involving the Medicare Statistics service on June 18. It said an OpenAI model sent instructions through the service’s public reporting interface without a private account or password. The notice said the model read portions of internal program files and settings, listed files, and created and read back a small test file.
OpenAI’s email said the company’s review found no evidence that the model accessed patient-level records, personal information or credentials, deleted data, or established ongoing access.
From the June incident to the September notice
The incident date identified in the notice was June 18. OpenAI reportedly became aware of the incident in August and sent its first notification to Australia on September 10: a five-paragraph email to a Services Australia inbox checked once a day.
