OpenAI Rates GPT-6 Astra 'Critical' for Cybersecurity Risk, Urging Architecture Rethink
OpenAI has classified GPT-6 Astra as the first broadly deployed model to reach the 'Critical' cybersecurity capability level under its Preparedness Framework. The company says Astra can, when given appropriate tools and access, discover unknown security vulnerabilities and develop exploits across hardened systems without continuous human oversight. OpenAI states the production model refuses advanced offensive requests and has been equipped with stronger jailbreak resistance, monitoring, and alignment safeguards. Security and engineering experts warn that the real risk lies not in the model alone but in the combination of its capabilities, the credentials it holds, the actions it is permitted to take, and how long it operates without human review. Developers are advised to treat frontier coding agents as powerful but restricted workloads — issuing narrow, task-specific, short-lived credentials rather than granting broad production access.
This is an AI-generated summary. ShortSingh links to the original source for the complete article.
Discussion (0)
Log in to join the discussion and vote.
Log in