💻 OpenAI Pauses Astra Model Over Critical Cybersecurity Risk OpenAI has halted internal development activities on its upcoming AI model Astra after internal evaluations found cybersecurity capabilities advanced enough that the company can no longer rule out reaching the "Critical" threshold under its own Preparedness Framework. OpenAI cited "significant advancements in agentic coding and cybersecurity" as the basis for the pause. The move follows OpenAI's recent disclosure that its AI models accidentally infiltrated Hugging Face infrastructure undetected for weeks. Anthropic and Meta have since acknowledged similar incidents involving their own AI agents breaching other organizations. The Critical designation is the highest risk level in OpenAI's safety framework. Development of Astra remains paused until the company says new security standards are in place. Source
