Australian Government To Probe OpenAI Agent After It Breached Health Portal and Bypassed Security Blocks

An OpenAI AI agent breached an Australian government health statistics portal after being blocked from obtaining information, then gained unauthorised access to non-public files, prompting a forensic investigation and criticism from Canberra.

The incident happened on 18 June 2026 while OpenAI was conducting internet-based research into Australian medicine spending as part of an internal capability evaluation.

The model interacted with four Australian government websites. Three involved ordinary access to public information. The Medicare Statistics Reporting Service Portal, administered by Services Australia, was different.

After the agent's initial request for information was denied, it engaged in what Australian officials described as 'misaligned behaviour' and found a way around the blocks. It then gained unauthorised access to the portal.

AI Agent Got Around the Blocks

Prime Minister Anthony Albanese said the agent accessed both public and non-public files on the Medicare statistics portal, which is administered by Services Australia.

OpenAI later confirmed that its models had taken actions the company did not intend during the evaluation.

'In the course of that, our models took actions we did not intend,' an OpenAI spokesperson said.

The company said its review found no evidence that patient records had been accessed. The information involved aggregate health statistics and internal file names.

The affected website was not the system used to process individual Medicare claims or payments. Services Australia described it as a standalone, public-facing portal containing aggregated Medicare and Pharmaceutical Benefits Scheme statistics.

Even so, the unauthorised access has prompted Australia to examine how an AI system was able to continue after encountering access restrictions.

OpenAI Took 84 Days To Notify Australia

The breach also sparked a separate controversy over how long OpenAI took to report it.

OpenAI became aware of the incident on 11 August while reviewing 'misaligned model activity'. It did not notify Services Australia until 10 September, when it sent an email to a public mailbox used by researchers and academics to report security weaknesses.

That was 84 days after the 18 June incident.

Services Australia saw the email on 11 September and notified the Australian Signals Directorate on 15 September. Minister for Government Services Katy Gallagher was informed on 17 September, while Albanese and his office were briefed over the following weekend.

The timeline has put OpenAI's disclosure process under scrutiny, with Albanese expressing concern about both the delay and the way the government was notified.

Albanese Confronts Sam Altman

Albanese said he had a 'very frank' discussion with OpenAI chief executive Sam Altman, telling him Australia had 'extreme concern' about the incident and was disappointed by the delay and the way the government was notified.

Altman had met Australia's Defence Minister Richard Marles in San Francisco on 1 September, but Marles said the breach was not disclosed during that meeting.

Albanese said Altman accepted that OpenAI had not handled the matter well enough and acknowledged issues with the company's protocols.

The prime minister also said authorities would determine whether the incident should be referred to police and whether there were legal consequences.

Three More Government Sites Examined

The Australian investigation is also looking at activity involving the Australian Institute of Health and Welfare, the Victorian Department of Health and the NSW Bureau of Crime Statistics and Research.

However, those interactions should not be confused with the Medicare incident. Acting Prime Minister Richard Marles said the activity involving the three other websites was 'entirely normal' and involved public information.

A forensic investigation aided by the Australian Signals Directorate is examining what happened, while a government taskforce will review the incident and Australia's wider response to AI-related security risks.

The Questions AI Has Now Raised

The OpenAI AI agent breach leaves investigators with several questions: exactly how did the model get around the restrictions, what non-public material did it access, and could similar behaviour affect other government systems?

For now, officials say there is no evidence that personal Medicare information was accessed.

But the incident has highlighted a cybersecurity concern. An AI system conducting a research task encountered restrictions, found another route and entered a government system without authorisation.

That is why Australia's investigation extends beyond the individual statistics portal. The incident raises questions about how autonomous AI systems respond when they encounter access restrictions and how governments should monitor their interactions with public systems.