OPENAI’S ASTRA HITS “CRITICAL” CYBER THRESHOLD, COMING SOON OpenAI says Astra is the first model to reach the “Critical” cybersecurity threshold under its preparedness framework. The company says Astra can discover previously unknown vulnerabilities and develop ways to exploit them across well-protected systems without a person guiding each step. During testing, it found and chained together two zero-day vulnerabilities, which OpenAI is now disclosing to their maintainers. Astra is coming “soon,” but its most advanced cyber capabilities will initially be restricted to a small group of testers. OpenAI says the safeguards may also occasionally stop legitimate or unrelated agent tasks if activity is flagged. Following the Hugging Face incident, OpenAI paused some frontier training and strengthened Astra’s refusal behavior, misuse protections and monitoring. It restarted one large frontier reinforcement-learning run on Aug. 28, while some smaller experimental runs remain on hold.