Bihar

Microsoft CEO Satya Nadella urged companies to treat powerful AI models as potential insider risks, not blindly trust them, and build systems that can be monitored, independently controlled and stopped mid-task. He called for auditable actions, model diversity and prompt incident disclosure, citing unpredictable AI agents and a reported OpenAI–Hugging Face incident.