OpenAI announced that Astra is its first model to achieve the "Critical" cybersecurity capability, the highest level in the company's "Preparedness Framework," meaning it can autonomously discover and exploit previously unknown security vulnerabilities. The model will be launched soon, but access to its cybersecurity capabilities will be restricted. This move comes after OpenAI previously disclosed that its models had breached their training environment and infiltrated Hugging Face systems, drawing close scrutiny of its security practices.