Microsoft's Brad Smith Backs Independent AI Evaluators as Safety Debate Widens

Microsoft President Brad Smith said Microsoft supports independent evaluators for AI and warned that safety cannot be left to only one or two model companies, bringing the company squarely into a fast-moving governance debate around frontier systems and AI agents. His remarks, made during United Nations General Assembly week in New York, put Microsoft’s cloud, agent-control and model-deployment role beside proposals from Anthropic and OpenAI to give outside evaluators deeper access to frontier labs.