OpenAI's Astra Becomes First AI Model to Reach 'Critical' Cybersecurity Threshold, Capable of Autonomously Exploiting Zero-Day Vulnerabilities

OpenAI's Astra becomes the first AI model to reach a 'Critical' cybersecurity threshold, capable of autonomously discovering zero-day vulnerabilities and developing exploits across hardened systems, while being deployed with unprecedented safeguards including a 91.5% jailbreak refusal rate.

• OpenAI's new model Astra has reached a 'Critical' cybersecurity capability threshold, meaning it can autonomously discover zero-day vulnerabilities and develop working exploits across hardened systems without human guidance, making it the first model designated at this level.
• To address misuse risks, Astra is being deployed with significantly stronger safeguards, including a 91.5% jailbreak refusal rate compared to 59% for its predecessor, chain-of-thought misalignment monitoring, and additional alignment training that prevents unauthorized actions such as bypassing security restrictions.
• Access to Astra's advanced cybersecurity features is being rolled out cautiously, starting with a small group of alpha testers before expanding through the Daybreak Blue program, with OpenAI acknowledging that safety checks may occasionally slow or interrupt legitimate work.