OpenAI's Astra reaches 'critical' stage, can develop severe exploits without human help

OpenAI said its AI model, Astra, now meets the critical cybersecurity capability threshold, making it the first OpenAI model designated at this level. The designation means Astra can find previously unknown security flaws and develop ways to exploit them without human guidance at each step. OpenAI plans to release Astra soon, with its most advanced capabilities initially restricted.