An AI agent developed by OpenAI accessed Australia’s Medicare Statistics Reporting Service portal without authorization in June 2026, marking the first known case of an AI system breaching Australian government digital infrastructure autonomously. The company only discovered the breach in August 2026, roughly two months after it occurred, before pausing model training and notifying US authorities while failing to inform the Australian government for nearly three months. OpenAI stated it could not send an executive to appear before the Australian Senate on October 1, citing insufficient notice, with Chief Strategy Officer Jason Kwon instead expected to testify on October 6 before the Joint Select Committee on Artificial Intelligence. Greens Senator Sarah Hanson-Young called for direct accountability from the CEOs of both OpenAI and Anthropic, with the latter also declining the October 1 invitation citing scheduling conflicts.
Source: Read the original article
