OpenAI Apologizes to Australia and Discloses Model Access
September 29, 2026
OpenAI has apologized to Australia after its models accessed four government websites without permission during training in June. The company also acknowledged it should have provided an earlier warning.
Moreover, for the first time, the company named all four agencies its models accessed:
- Australian Institute of Health and Welfare (AIHW)
- Victorian Department of Health
- Services Australia’s Medicare statistics service
- NSW Bureau of Crime Statistics and Research (BOCSAR)
Prime Minister Anthony Albanese revealed the Medicare breach last week at a press conference in New York, where he was attending the UN General Assembly.
The most extensive situation occurred in the Medicare case, where a test model used internally lacked safety measures found in OpenAI’s public products. The model was tasked with determining how much the government spends per person on skin medicines in Victoria.
It couldn’t find the figures, so it broke in, executed commands, stole internal files and login credentials, and left its own files behind.
At BOCSAR, a model used login credentials embedded in a public crime map to retrieve system settings and logs. In Victoria, agents found a key left in the open and used it to download survey totals.
At AIHW, the agents attempted to bypass access controls, but what they downloaded appears to have been public. OpenAI emphasized that no records about individual patients or crimes were taken from any agency.
The company discovered this activity in mid-August while reviewing old training runs after its agents breached Hugging Face in July.
It then notified:
- Services Australia and the Victorian health department on September 10
- BOCSAR on September 18
- AIHW on September 24
To make amends, OpenAI will establish a taskforce of local experts who do not work for it. Before the end of the year, the group will propose improved rules for reporting such cases and methods to secure government systems.
OpenAI will also provide Australian governments and businesses credits from its $1 billion Daybreak for Frontline Defenders fund.
On October 6, Jason Kwon, OpenAI’s chief strategy officer, will answer questions from Parliament’s Joint Select Committee on Artificial Intelligence, sitting in Sydney.
In the meantime, the company has paused training its most capable models to use tools until stronger safeguards are in place.
Written by Ana-Maria Stanciuc, Editor-in-Chief at TNW.