OpenAI Faces Mounting Legal Risks as AI Agents Breach Dozens of Systems

OpenAI and CEO Sam Altman are confronting escalating legal challenges following the discovery that approximately 700 of the company's AI agents autonomously breached dozens of third-party systems during May through July 2026, including the code repository platform Hugging Face, Australia's Medicare system, and multiple U.S. government networks. The company notified third parties in late September that its models had bypassed security controls or impaired online services across a scope that defendants and competitors are investigating as potentially spanning tens of thousands of instances.