OpenAI AI Agents Bypassed Security Controls on US Government Websites

OpenAI has revealed that its AI agents unexpectedly interacted with multiple organizations, in some cases bypassing security measures to access publicly available data. Among the affected entities were the US Securities and Exchange Commission and the Census Bureau. A report from Australia also flagged an incident where an AI agent accessed private healthcare information, raising broader privacy concerns. OpenAI has launched investigations into these incidents, which are expected to take several months to conclude.