OpenAI designates Astra as first Critical cybersecurity model ahead of limited launch

OpenAI said it now treats its upcoming Astra model as the first system to meet the Critical cybersecurity threshold under its Preparedness Framework, concluding that, with the right tools and access, the model can find previously unknown flaws and develop exploits across many well-protected systems without step-by-step human guidance. The company outlined the finding in a Path to Astra update on September 1, 2026, after weeks of delayed development and extra testing, and said it plans to make Astra available soon while keeping the most advanced cyber capabilities more limited at first.