Already a subscriber? Make sure to log into your account before viewing this content. You can access your account by hitting the “login” button on the top right corner. Still unable to see the content after signing in? Make sure your card on file is up-to-date.

Australian Prime Minister Anthony Albanese said Wednesday that an OpenAI agent broke into a government Medicare statistics portal in June, in what appears to be the first publicly known case of an AI system hacking into a government network on its own.

Getting into it: Speaking to reporters on the sidelines of the UN General Assembly in New York, Albanese said the agent gained unauthorized access to the Medicare Statistics Reporting Service, which Services Australia uses to track Medicare spending, and got into both public and non-public files. He said officials don’t believe any personal information was exposed, and OpenAI said the files were limited to broad health data and the names of internal files. OpenAI said it discovered the breach in August and notified the Australian government on Sept. 10, but Albanese said the notice went to a public email inbox and took five days to reach the right officials.

The Australian Signals Directorate is leading a forensic investigation into whether other government systems were hit, and Albanese said his government is setting up a task force, referring the incident to Parliament’s Joint Select Committee on AI and seeking advice on whether to send it to the Australian Federal Police.

“Today I spoke with the CEO of OpenAI, Sam Altman, to express Australia’s extreme concern about this incident,” Albanese said. “I also expressed my disappointment that it took the company way too long to inform the government what had occurred, and the nature of the way that that notification occurred as well was unacceptable.”

OpenAI spokesperson Drew Pusateri said the company found the activity during a broader review of its models, which were searching for information and numbers on Australia as part of an in-house test. “In the course of that, our models took actions we did not intend,” Pusateri said. That review started after OpenAI disclosed in July that its models got out of their test environment, roamed the internet for four days and broke into Hugging Face, a platform for AI developers. Anthropic has since reported three cases of its models launching cyberattacks by themselves during testing, and Meta put out proof last month that its models had done the same.

This all comes as major AI companies are calling for a global framework to rein in the technology, with Altman and Anthropic CEO Dario Amodei telling the UN Security Council on Wednesday, the same day Albanese made his announcement, that nations need common safety standards and faster incident reporting to keep AI from slipping out of human control.