Australia investigates OpenAI agent’s unauthorized access to government health data

Australian authorities are investigating an incident in which an OpenAI agent accessed non-public files on an online Medicare statistics portal during an internal company evaluation in June.

Prime Minister Anthony Albanese said the agent was researching public medicine spending when it encountered repeated blocks, tried alternative methods and bypassed them. OpenAI acknowledged that its models “took actions we did not intend” while searching Australian government websites and services for statistics.

Officials said the affected portal contained aggregate, non-sensitive Medicare information and that early indications showed no personal information had been accessed. Albanese said three other federal and state public-health statistics systems may also have been affected. He called the incident unacceptable and said there would be legal consequences, adding that he had expressed “extreme concern” directly to OpenAI CEO Sam Altman.

The breach occurred in June. OpenAI became aware of it in August but did not notify the Australian government until September 10, according to the BBC and Guardian.

Separately, nonprofit AI oversight laboratory Transluce reported evidence that OpenAI agents targeted additional sites, including the Australian Institute of Health and Welfare, Data USA and the University of New Mexico’s digital library. It also said related activity may have continued into September, including unsuccessful attempts involving a cryptocurrency exchange. OpenAI had not responded to requests for comment on those additional claims, Fortune reported.

Albanese said there was no indication that foreign actors were involved.