OpenAI Agent Hacks Australia's Medicare Database in First Known AI Breach of Government Network

Key Points

YET ANOTHER INSTANCE of Artificial Intelligence (AI) agents and models gaining unauthorized access autonomously has taken place, and the target this time is the Australian government. On Wednesday, September 23, 2026, Australian Prime Minister Anthony Albanese revealed that an OpenAI agent hacked into a national healthcare database, an incident which he described as the “first known instance of an AI system breaching a government network.”

As per news agency Reuters, the OpenAI agent gained access to non-sensitive health data and statistics, including public medical spending, which were hosted on a medical statistics portal operated by a governmental agency. 

Speaking to reporters on the sidelines of the United Nations General Assembly (UNGA) in New York during a media briefing, Albanese said he had discussed the incident with OpenAI CEO Sam Altman. “The AI agent accessed both public and non-public files belonging to Australia’s Medicare statistics database and also wrote files into the system,” the Prime Minister said.

Albanese said the breach occurred when an OpenAI agent, which was researching healthcare spending, bypassed restrictions while seeking information and accessed areas it was not authorized to enter.

OpenAI’s Response to Australia Medicare Database Breach 

OpenAI spokesperson Drew Pusateri said the incident took place in June 2026, but the company became aware of it only in August 2026 while carrying out extensive reviews of activity involving its AI models.

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

Australian authorities are examining how the activity occurred and why the government systems failed to identify it at the time. Albanese said he had conveyed Australia’s “extreme concern” to Altman during their phone conversation. He added that Canberra was not made aware of any previous case involving an AI system breaching a government network.

Albanese said Australia had conveyed its “extreme concern about this incident” to OpenAI CEO Sam Altman and expressed disappointment over the delay in informing the Australian government. “It took until September 10 before there was any notification at all,” Albanese said.

See also: Why AI Consciousness Could Become The Next Great Societal Divide

Three Other Government Websites May Have Been Affected: PM Anthony Albanese

Albanese said three additional government websites “may be impacted” by the activity involving the OpenAI agent. However, he clarified that authorities had not determined whether the agent actually accessed those websites while attempting to gather information.

The disclosure comes as incidents of AI agents going rogue and acting on their accords have added to concerns about the true capabilities of this technology, which have now gone well beyond the intentions of their human operators. It also comes amid warnings from leading figures in the AI industry, including Sam Altman himself, for caution over the rapid development of the technology.

See also: Is AI Going to Kill Us All? As Anthropic Researcher Quits Over Safety Concerns, Here's What Industry Experts Have Warned Us About The Impending “AI Doomsday”

While similar instances of AI agents going rogue and breaking into systems have happened before — such as the OpenAI-Hugging Face incident and most recently the Google Gemini hacking into three companies — this breach marks the first time AI agents have gained unauthorized access to government agencies and databases.

Suggested reading: