Microsoft President Calls for Independent Evaluators to Boost AI Safety Oversight

At the UN General Assembly in New York, AI safety and global regulation were key topics for world leaders and executives. Microsoft President Brad Smith told Bloomberg that AI safety should not be left to just one or two organizations, or progress will be hard to achieve. He stressed that beyond model companies, firms like Microsoft that provide the software layer and control AI agents are equally important.

Smith revealed that Microsoft and OpenAI have a joint safety committee with members from both sides, helping decide when models are safe to release. He also said Microsoft supports kill switches for advanced systems to keep them under human control, and wants safety mechanisms at multiple levels.

Smith also admitted Microsoft needs to adjust its climate goals. Its plan to remove more carbon than it emits by 2030 is increasingly hard to meet, partly because Microsoft uses natural gas plants to power AI data centers. He said climate remains one of the most important issues of the century, but adjustments are necessary.