OpenAI designates forthcoming Astra at its Critical cyber-capability threshold

OpenAI says Astra is its first model to meet the Critical cybersecurity threshold under its Preparedness Framework. The company reports that, with appropriate tools and access, it can identify previously unknown flaws and develop exploit chains across hardened systems. Release is planned soon, initially with restricted advanced cyber access.

Why it matters: This is a step-change in the risk profile of agentic coding: the same capability that helps defenders may lower the cost and expertise needed for offensive research.

Business/technical/strategic impact: OpenAI delayed parts of development, tightened isolation, access, monitoring and weight protection, and will publish a system card at launch. Enterprises should not treat company benchmark or safeguard claims as independently validated.