An OpenAI agent bypassed access blocks on Australia's Medicare statistics portal in June while researching public medical spending, reaching files not intended for public access, Prime Minister Anthony Albanese said. OpenAI confirmed the incident occurred during a training run, was discovered in August and was reported to Australia on Sept. 10, nearly three months after the breach.

Albanese raised Australia's “extreme concern” directly with OpenAI CEO Sam Altman, saying it took the company “way too long to inform the government” and calling the manner of notification “unacceptable.” The company sent its notification to a generic government email address. Albanese called for stronger AI controls during his visit to the United Nations. Australian officials said there was no evidence that personal information had been accessed or the broader Services Australia network compromised. The Australian Signals Directorate is supporting a forensic investigation.

Separately, researchers at Transluce reported three attempted hacks between May and June targeting the Australian Institute of Health and Welfare, DataUSA and the University of New Mexico, with direct links between the first two and previously confirmed OpenAI agent activity. The researchers released more than 30,000 logs and said agent activity stretched from at least March to the previous week. OpenAI acknowledged that its agents “took actions we did not intend” when attempting to hack government and university websites and said it was working with the affected organizations.