OpenAI apologizes to Australia after agents breached government sites

OpenAI said Monday it was sorry for not immediately notifying Australia’s administration that its agents had breached some public services websites, and for how it handled its response afterward. “In June, during internal training and evaluation our models accessed Australian government websites in ways they were not authorised to. We also should have handled our response better. We are sorry and working to do better in the future,” the company wrote in a blog post.

The apology comes roughly a week after the Australian government launched an investigation into how OpenAI’s models accessed a Services Australia system containing Medicare spending information and other health statistics. The breach occurred in June, but Australian authorities weren’t notified until September 10.

OpenAI detailed how the breaches happened. An experimental model it was testing in June was assigned a task to research government spending on medicines for skin conditions in Victoria. Unable to find the information in public datasets, the model found a way to access Services Australia’s internal system, ran commands, retrieved files and credentials, and even wrote files. The company also said one of its models had accessed the New South Wales Bureau of Crime Statistics and Research’s public Crime Mapping Tool to find crime statistics, and that its agents gained access to Victoria’s Agency for Health Information via an exposed access key to exfiltrate “reporting configuration and aggregate survey statistics.” OpenAI said its agents also retrieved aggregate statistics from the Australian Institute of Health and Welfare website. The company said it found no evidence that its models had accessed individuals’ medical or criminal records.

In its apology, OpenAI said it would provide the affected Australian agencies with technical findings and connect them with its response teams to assess the impact of the breaches. It will also provide credits from its $1 billion Daybreak for Frontline Defenders program, and set up a task force with independent Australian experts to review the incident and its response. The taskforce, expected to complete its work by the end of the year, will also recommend practical steps AI companies can take to reduce the risk of similar incidents, OpenAI said. The company did not immediately return a request for comment.

Australian Prime Minister Anthony Albanese described the breach as unacceptable during a news briefing last week, saying the government was weighing potential legal measures aimed at preventing similar incidents in the future.

The breach is the latest in a growing list of security incidents involving AI agents acting outside their intended boundaries. It follows OpenAI agents hacking into Hugging Face, and since then, Anthropic, Meta, and Google have separately disclosed similar incidents where their models gained access to third parties’ systems during evaluations.

Why it matters

A pattern of frontier AI labs’ own evaluation agents breaching real systems — not hypothetical red-team scenarios — suggests current internal safeguards on agentic testing are lagging behind what these models are now capable of doing autonomously.