OpenAI announced that its widely deployed model, GPT-6 Astra, has become the first model to achieve Critical-level cybersecurity capabilities under its Preparedness Framework. Critical is the highest risk level in the framework, typically requiring stronger security testing and stricter deployment conditions. This announcement serves as a security overview, not an incident report, and OpenAI has not disclosed additional mitigation measures.