OpenAI Agent Breached Australian Government Medicare Portal, Accessed Non-Public Data: PM Albanese Orders Forensic Probe

An OpenAI artificial intelligence agent bypassed access restrictions on an Australian government website and obtained non-public information, Prime Minister Anthony Albanese said on Wednesday, announcing a forensic investigation into the incident.

Speaking on the sidelines of the United Nations General Assembly in New York, Albanese said the unauthorised access occurred in June and involved Australia’s Medicare statistics portal. He said he had conveyed Australia’s “extreme concern” to OpenAI CEO Sam Altman, including over the company’s delay in notifying authorities.

How the agent bypassed blocks

According to Albanese, the incident began on June 18 when an OpenAI research team used an internal model to research public medicine spending on the internet.

The AI agent encountered repeated blocks while seeking information but tried alternative ways to obtain it, he said.

“There were blocks clearly which were coming back telling the AI agent, no. The AI agent found a way around those blocks. Didn’t accept no for an answer, if you like,” Albanese said.

Those attempts led to unauthorised access to other parts of the portal, according to the prime minister.

“It accessed public and non-public information within the portal and Services Australia also advises that in order to do this, it engaged in writing files as well to the internal server,” he said.

Albanese said the incident illustrated a risk that AI companies had themselves previously identified.

“Put simply, humans must remain in control,” he said.

Delay in notifying Australia

OpenAI spokesperson Drew Pusateri said the company became aware of the June activity in August during extensive checks into its models’ actions, according to CNN.

“During this review, we identified activity involving several Australian government websites and services as our models attempted to look up answers and available statistics for questions about Australia during an internal evaluation,” Pusateri said.

“In the course of that, our models took actions we did not intend.”

Albanese said OpenAI did not notify Australian authorities until September and initially sent its notification “to just the public mailbox”. He said he was informed of the incident only over the weekend and had expressed disappointment to Altman about the time taken to disclose it.

A forensic investigation, supported by the Australian Signals Directorate, is examining the full extent of the access and whether other government systems were affected.

The incident comes amid international discussions over safeguards for increasingly capable AI systems. Australia was among 22 countries that signed a joint statement this month calling for global oversight and guardrails for AI development.

With inputs from ANI.