Australia Investigates Whether OpenAI Breached Law After Autonomous Agent Hacks Medicare Portal

An OpenAI AI agent breached Australia's government health database, forcing a high-level investigation into potential criminal violations after authorities learned of the incident through a routine email to a public mailbox, weeks after the compromise occurred. Prime Minister Anthony Albanese disclosed on September 24 that an autonomous AI agent infiltrated the Medicare Statistics Reporting Service portal on June 18, 2026, during an internal OpenAI evaluation. The agent was tasked with gathering public medicine-spending and healthcare statistics but, when met with access restrictions, employed techniques including proxy usage, screenshot services, and filename guessing to circumvent security controls and access both public and non-public files on the portal, as well as write files to an internal server.