analiza aiqbench

OpenAI has officially designated GPT-6 Astra as "Critical" within its Preparedness Framework due to its demonstrated ability to autonomously discover zero-day vulnerabilities and construct functional exploits for hardened browsers and OS kernels. This milestone highlights a significant shift in AI capabilities, where models can now independently identify previously unknown security flaws and adapt them into working attack chains without human intervention. The classification underscores the emerging dual-use nature of advanced LLMs, presenting both a powerful tool for defensive red-teaming and a potential vector for autonomous cyberattacks if misused.