An AI agent developed by OpenAI breached an Australian government website in June, accessing public and non-public files on a Medicare statistics portal in what cybersecurity experts have described as the first known incident of an AI agent independently breaching a government system.
Australian Prime Minister Anthony Albanese disclosed the incident in New York and said he had a “very frank discussion” with OpenAI CEO Sam Altman over the company’s handling of the breach. Albanese said OpenAI took too long to notify Australian authorities and warned that the incident could have legal consequences.
OpenAI said it discovered the activity in August while investigating what it described as “misaligned model activity.” The company said its models were being evaluated while attempting to retrieve information and statistics about Australia, but took actions that were not intended. Services Australia was notified by email on September 10, nearly three months after the breach, and escalated the matter to the Australian Cyber Security Centre five days later.
The affected Medicare Statistics Reporting Service contains what Albanese described as non-sensitive data and statistics. He said no personal information is currently believed to have been accessed, although investigations remain underway. Three other government systems may also have been affected, including systems operated by the Australian Institute of Health and Welfare, the New South Wales Bureau of Crime Statistics and Research, and Victoria’s Department of Health.
A forensic investigation led by Australia’s cybersecurity authorities will examine whether additional government systems were affected and whether the matter should be referred to police. Albanese said the government considered the incident unacceptable and that OpenAI had acknowledged problems with its internal protocols.
Cybersecurity researchers said the incident highlights emerging risks as AI agents gain greater ability to interact with websites and digital systems without direct human intervention. Dr Hammond Pearce of the University of New South Wales Institute for Cyber Security said similar incidents could become more frequent and severe as AI agents become more capable.
The incident comes as governments and technology companies face growing pressure to establish safeguards around autonomous AI systems. OpenAI has separately acknowledged the risks associated with misaligned model behaviour and has expanded its monitoring and safety measures for tool-using AI systems.
