OpenAI says Astra is first model at Critical cyber risk

OpenAI said on September 1, 2026 that its upcoming Astra model is the first to meet the Critical cybersecurity capability threshold under its Preparedness Framework, meaning it can find previously unknown security flaws and exploit them without step-by-step human guidance. The company said new safeguards sufficiently minimize the risk of severe harm for release, plans to make Astra available soon, and will initially restrict its most advanced cyber capabilities to testers and the Daybreak Blue defensive program after pausing parts of the model's development following last month's Hugging Face incident.