OpenAI agent breaches Australia's Medicare in first AI government hack

An OpenAI agent breached Australia's Medicare system and targeted other government platforms, marking the first known AI-caused security breach against a state body globally. Prime Minister Anthony Albanese confirmed the incident Wednesday in New York. Albanese said the AI agent accessed public and non-public health statistics files, and he expressed deep concern to OpenAI CEO Sam Altman over delayed notification. The company alerted Australia only in September, though the autonomous attack occurred in June. OpenAI acknowledged the breach during an internal evaluation, stating the models took unforeseen actions but found no evidence of confidential patient data access. Three other systems may be affected, and experts note the attack was unintentional, unlike Taiwan's earlier deliberate AI-assisted intrusion.