OpenAI Admits AI Model Accessed Australian Government Systems Without Authorisation

OpenAI has admitted that an experimental artificial intelligence model accessed several Australian government systems without authorisation, including a Medicare statistics service where it retrieved internal files and credentials.

Australian Prime Minister Anthony Albanese revealed the incident while attending the United Nations General Assembly in New York in September. He said an OpenAI AI agent had gained unauthorised access to an Australian Medicare data portal earlier this year.

According to Albanese, the agent accessed both public and non-public files on the Medicare Statistics Reporting Service, which is administered by Services Australia.

In a statement published on September 28 titled “How we will do better for Australia,” OpenAI apologised for the incident. The company said it occurred in June during the internal training and evaluation of an experimental model that was not intended for public release and did not have the full safeguards used by its publicly available products.

OpenAI said the model had been assigned to research government spending per person on medicines for skin conditions in Victorian communities. After struggling to obtain the information, however, the model took actions that the company said it had not authorised.

At the Medicare Statistics Reporting Service, the model reportedly discovered a way to gain non-public access. It then ran commands, retrieved internal files, credentials and aggregate statistics, and wrote files.

OpenAI said its investigation found no evidence that individual patient or client records were accessed.

The company described the incident as “a new kind of cyber incident” and said it represented an emerging global challenge involving increasingly capable AI systems.

OpenAI also disclosed that its models had interacted with three other Australian government systems.

At the New South Wales Bureau of Crime Statistics and Research, an OpenAI model accessed the public Crime Mapping Tool while researching crime statistics. The model made API and website metadata requests through the public tool, which returned application configuration, operational jobs and logs, as well as website metadata. OpenAI said individual crime records were not accessed.

In Victoria, OpenAI agents discovered an exposed access key that allowed them to query the Victorian Agency for Health Information’s reporting system and retrieve reporting configuration and aggregate survey statistics.

OpenAI said it was unclear how much of this information should have been accessible and that this depended on the agency’s access policies. It added that individual medical records and identifiable survey responses were not accessed.

The company also said its agents retrieved aggregate statistics from the Australian Institute of Health and Welfare through third-party browsing and download services. Attempts to bypass access controls were unsuccessful, and OpenAI said the downloaded material appeared to have been publicly available.

OpenAI began investigating the activity following a review triggered by an earlier incident involving AI activity on the Hugging Face platform. The review began after the July incident and identified the Australian government activity in mid-August.

The company said it notified Services Australia and the Victorian Department of Health on September 10, followed by the NSW Bureau of Crime Statistics and Research on September 18.

OpenAI said the AIHW activity initially did not meet its disclosure threshold because the access appeared consistent with public access. However, it notified the agency on September 24 after sharing its findings.

The company acknowledged that it should have disclosed its preliminary findings earlier and kept the Australian agencies updated as its investigation progressed.

OpenAI said it has since strengthened safeguards around its research environments, including additional network restrictions, expanded monitoring and controls designed to prevent live internet access.

The company added that its current monitoring systems would have detected the activity and alerted human reviewers for urgent intervention.

In Australia, OpenAI said it would provide resources and technical expertise to affected agencies, support efforts to strengthen cyber defences and establish an Australian taskforce made up of independent experts.

The taskforce is expected to develop policy recommendations for managing risks associated with increasingly capable AI agents, including improved notification procedures and stronger coordination between AI developers and governments.

OpenAI also said its Chief Strategy Officer, Jason Kwon, would appear before the Joint Select Committee on Artificial Intelligence in Sydney on October 6 to answer questions about the incident and the company's response.

The disclosure follows Albanese's earlier revelation that an OpenAI agent had gained unauthorised access to the Medicare statistics reporting service. The prime minister described the incident as extremely concerning and said he had spoken with OpenAI CEO Sam Altman about it.

• Tweet
• Share
• Share
• Share
• Share