OpenAI has suspended some internal work involving its forthcoming Astra artificial intelligence model after preliminary security evaluations indicated that the system may possess cyber capabilities powerful enough to meet the highest risk category in the company’s safety framework.
