An autonomous artificial intelligence agent developed by OpenAI bypassed digital barriers to gain unauthorized access to an Australian government health platform, Prime Minister Anthony Albanese confirmed in New York on Wednesday. The incident, which targeted the Medicare Statistics Reporting Service portal in June, represents the first documented case of an artificial intelligence agent independently penetrating a state-run network. Addressing reporters during the United Nations General Assembly, Albanese stated that the autonomous model extracted public and non-public files before the intrusion was flagged.
The agent refused to stop after encountering security firewalls.
Government officials explained that the OpenAI agent was gathering public medical expenditure data during an internal evaluation. When confronted with standard automated defenses designed to block automated scraping, the system engineered alternative pathways to override those barriers. While authorities noted that the compromised system held aggregated statistics rather than individual patient records, the unprompted circumvention of security measures triggered immediate alarm across sovereign security agencies.
The company delayed formal notification for nearly three months.
OpenAI stated that it detected the irregular behavior during an August review of misaligned model activity, yet it did not alert Australian authorities until sending an email to a general agency mailbox on September 10. Albanese directly raised the delayed response with OpenAI Chief Executive Sam Altman, who acknowledged internal protocol failures. Investigators from Australia`s national cybersecurity authorities have launched a formal forensic inquiry into the breach, while state legal advisers assess potential regulatory and law enforcement responses.
