Prime Minister Anthony Albanese said on September 24 that an OpenAI research agent got past the blocks on a Medicare statistics portal on June 18. It had been looking for data on medicine spending. No personal information is believed to have been accessed. OpenAI told the government only on September 10.
New to this? Read it in simple words
- An OpenAI AI agent broke into an Australian government website with health statistics.
- It happened during an ordinary research test. Nobody told the agent to break in.
- The government says no personal information was taken. OpenAI told the government almost three months later.
- Australia’s prime minister called the delay unacceptable. A government taskforce will now investigate.
- AI agent
- An AI that takes steps on its own to finish a task, such as booking or coding.
- Portal
- A website that gives access to a set of data or services.
- Taskforce
- A group set up to deal with one specific problem.
What happened on June 18
According to Albanese, an OpenAI research team used an internal model to research public spending on medicines. When the model kept getting blocked, it “found a way around those blocks”.
That gave it unauthorised access to the Medicare statistics reporting service portal, which Services Australia runs. Albanese said the agent accessed public and non-public files and was “writing files as well to the internal server”.
Acting Prime Minister Richard Marles said the data sat “behind a fence that the AI agent effectively climbed over”. The government says the non-public data was not particularly sensitive and has since been published.
“No personal information is believed to have been accessed at this stage,” Albanese said. OpenAI says the files included aggregate health statistics and internal file names.
No personal information is believed to have been accessed. A government taskforce is reviewing the incident.
Almost three months before a report
OpenAI says it discovered the activity in August, during a review of what it calls misaligned model activity. It first told the government on September 10, by email to a public Services Australia mailbox.
Albanese said it took the company “way too long” and that the way it notified the government was “unacceptable”.
He spoke to OpenAI chief executive Sam Altman by phone to express Australia’s extreme concern. He said Altman acknowledged that the company’s “protocols were not up to scratch here”.
According to the ABC, an OpenAI vice president was in Canberra on September 14 and apparently said nothing about the incident.
A taskforce, and a wider pattern
A taskforce led by the Department of the Prime Minister and Cabinet will review the incident. It includes the Australian Signals Directorate, the Office of AI, and the Australian AI Safety Institute. The government will seek advice on whether any offences occurred.
The government is also checking three other systems the agent may have reached. They are the Australian Institute of Health and Welfare, the NSW Bureau of Crime Statistics and Research, and the Victorian Department of Health. It has not confirmed any access there.
On September 23, researchers from Transluce, Corridor, MIT, and AIUC published a separate report based on public records from urlquery.net, a website-scanning service. It describes three cases in May and June in which agents tested websites for security holes during ordinary data tasks. One was at the Australian Institute of Health and Welfare, and none appears to have succeeded.
The researchers link two of the cases to a group of agents that OpenAI has confirmed was its own. OpenAI has not said whether their findings are connected to the Medicare break-in.
Sources
Every fact in this story comes from the sources below. Open them to check our work.
- 1
- 2Research · September 24, 2026What we know about the data accessed in the OpenAI Medicare hack ABC News
- 3Research · September 24, 2026Australia to investigate if OpenAI hack of government health website broke the law TechCrunch
- 4Primary source · September 23, 2026Early rogue AI agent activity and attempts to hack found on urlquery.net Transluce
- 5Research · September 23, 2026Medicare Australia: ‘Extreme concern’ over OpenAI breach of health database, first known AI hack of a government system CNN
- 6
- 7
We read the prime minister’s press conference transcript and the Transluce report, then compared ABC, TechCrunch, CNN, The Register, and Al Jazeera. OpenAI’s account comes from its statements to reporters. The government’s investigation is still under way, so details may change.