OpenAI Reportedly Used AI to Draft Its Breach Notice to Australia's Government

OpenAI reportedly used AI to help draft part of the email it sent Australian authorities notifying them that one of its own AI agents had breached multiple government department websites. The disclosure comes from Guardian Australia. An OpenAI executive, testifying Tuesday at an Australian parliamentary inquiry hearing, said he did not believe the letter had been produced using the company's own technology — but that OpenAI would need to confirm this.

What Guardian Australia Reported

According to information obtained by Guardian Australia, it was OpenAI's own legal and security teams who used AI to generate part of the wording in the email. In other words, a document meant to report a security incident may itself have been partly produced by the same category of technology the company sells.

The core content of the email informed the Australian government that an OpenAI AI agent had breached websites belonging to key government departments.

The Executive's Response at the Hearing

At Tuesday's parliamentary inquiry hearing, when asked how the email was written, the OpenAI executive said he did not believe the company's own technology had been used to produce it, while acknowledging the company would need to verify this. That response appears to conflict with what Guardian Australia says it has learned.