Microsoft President Brad Smith said the company backs independent safety-assessment bodies for AI deployments, arguing AI safety should not depend on a small number of model firms. He acknowledged OpenAI and Anthropic as important but said software firms that control and monitor AI agents are equally critical. Microsoft participates in safety assessments via a committee set up with OpenAI and supports human-operated shutdown mechanisms for advanced AI systems. Smith said AI safety requires multi

Microsoft President Brad Smith said the company backs independent safety-assessment bodies for AI deployments, arguing AI safety should not depend on a small number of model firms. He acknowledged OpenAI and Anthropic as important but said software firms that control and monitor AI agents are equally critical. Microsoft participates in safety assessments via a committee set up with OpenAI and supports human-operated shutdown mechanisms for advanced AI systems. Smith said AI safety requires multi-layered defenses and that key controls should not be centralized. He confirmed Microsoft still plans to deploy 200 MW of compute at a UAE AI data center and views the Middle East as a timely investment destination. On its carbon-negative-by-2030 target, he said the company may need to adjust its previously set pathway.