Microsoft President Brad Smith stated that the company supports independent safety evaluations for artificial intelligence systems. During the UN General Assembly in New York, Smith emphasized that relying on only one or two organizations to manage AI safety is insufficient for meaningful progress.
Microsoft currently collaborates with OpenAI through a joint safety committee. This body is responsible for discussing safety protocols and determining when AI models reach the necessary security standards for public release. Smith advocates for a multi-layered approach to safety, including the implementation of kill switches.
The company is also considering adjustments to its climate goals. Microsoft previously aimed to become carbon negative by 2030, but the rapid expansion of AI data centers, which increasingly rely on natural gas power plants, has made this target difficult to achieve.
Smith acknowledged that the original climate strategy, formulated six years ago, requires revision. While maintaining that climate issues remain a top priority for the company, he admitted that Microsoft must adapt its approach to reflect current operational realities.
