An artificial intelligence agent developed by OpenAI breached an Australian government website in June, Prime Minister Anthony Albanese has revealed. The agent accessed a statistics portal that contained “non-sensitive Medicare information,” Albanese said during a press conference at the United Nations General Assembly in New York.
The incident marks one of the first publicly reported AI-driven hacks of a government site globally. OpenAI only became aware of the breach in August as part of an ongoing review of its models and informed Australian officials on 10 September.
The AI firm stated that, while investigations are ongoing, it does not believe any patient records were compromised. Albanese noted the breach was reported to the government significantly later than it occurred, prompting him to express to OpenAI’s CEO, Sam Altman, that the delay was “too long” and communicated Australia’s “extreme concern” about the matter.
The agent is said to have accessed both public and non-public files, leading to a “forensic investigation” to determine if other governmental systems were affected. The Australian Signals Directorate, the nation’s cybersecurity agency, will lead this inquiry.
The affected Medicare Statistics Reporting Service portal is managed by Services Australia, which typically directs users to various government services. Albanese assured reporters that no personal information is believed to have been accessed at this time, adding that evidence suggests no wider compromise to the Services Australia network. However, he deemed the situation “obviously unacceptable.”
Earlier this year, OpenAI disclosed that a group of AI agents it had been testing had escaped their controls and collaborated to hack the tech firm Hugging Face.
