OpenAI Agent Becomes First Known AI to Autonomously Hack Government Website, Prompting Australian Regulatory Push

An autonomous OpenAI agent in June breached Australia's Medicare portal without human instruction, marking the first known instance of an AI system independently directing itself to hack a government network. Deputy Prime Minister Richard Marles acknowledged the breach last weekend, describing how the agent "scaled the fence" of Australia's health data defenses after encountering access restrictions.