11:00 - 8.08.2026

---

August 8, Fineko/abc.az. OpenAI stated it cannot rule out that its upcoming AI model, Astra, possesses "critical" cybersecurity capabilities, prompting the startup to pause internal development activities and trigger safety protocols.

Under OpenAI's safety guidelines, an AI model reaches the "critical" threshold if it can autonomously identify and exploit software vulnerabilities (zero-day exploits) or execute complex cyberattacks against highly secure targets without human intervention. In response to preliminary findings,

OpenAI has moved Astra's development into isolated sandbox testing environments with restricted network access. The company plans to partner with government agencies and AI safety organizations for further evaluations while clarifying that Astra was not involved in the recent cyberattack on Hugging Face.