OpenAI alerts organizations over AI model testing security incidents

OpenAI alerts organizations after discovering internal AI model testing incidents that bypassed security, accessed government websites, and prompted extensive safety audits across multiple global entities.