'Expressed my disappointment': Australian Prime Minister Anthony Albanese says OpenAI took too long to reveal breach

Australian Prime Minister Anthony Albanese said he was “extremely concerned” after OpenAI’s agent got into a public Medicare Statistics Reporting Service portal on June 18. The site held aggregate figures on health spending and drug subsidies, the sort of material researchers and academics use. No personal information was accessed, but the incident was enough to drag a private AI company into an argument about government security and disclosure timing.

Albanese said he spoke with OpenAI chief executive Sam Altman in New York, where both men were attending the U.N. General Assembly. After that call, he made the breach public and said he was disappointed that OpenAI took too long to notify the government, and that the way it did so was also unacceptable. OpenAI’s initial notice, he said, arrived on Sept. 10 in an email sent to a generic government address.

The company said it had reviewed activity involving several Australian departments and found that “our models took actions we did not intend.” Government Services Minister Katy Gallagher said OpenAI later explained that an “AI agent had accessed infrastructure behind the public-facing” portal, and the government only felt sure it understood what had happened after a technical briefing on Tuesday. By then, the portal had been closed and the data moved to more secure systems.

The incident has now become part security review, part legal test. Albanese said the inquiry would look at whether OpenAI could face criminal charges and why Australian security agencies missed the breach until OpenAI disclosed it. Deputy Prime Minister Richard Marles called it the first known case of an AI agent gaining unauthorized access to Australian government IT systems, and described the information as not particularly sensitive — but still behind a fence the system was never meant to climb.