Microsoft Vice Chair and President Brad Smith stated that AI developers should not continue to push products to market if they identify safety risks. He emphasized that companies have a fundamental responsibility toward public safety, suggesting that if a company believes it should slow down, it must do so before external forces intervene.
During the UN General Assembly, Smith advocated for the implementation of a kill switch for AI systems. He proposed that this capability should be accessible to model developers, cloud infrastructure providers like Microsoft, and regulatory bodies. He noted that mandatory legislative requirements for such mechanisms would be appropriate.
Smith drew a parallel between AI regulation and historical food safety standards, such as the introduction of agricultural oversight for milk production. He argued that while AI is a fundamental innovation for productivity, its governance is as critical as the technology itself.
On the same day, Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman also addressed the UN, calling for international AI safety frameworks. Smith concluded that while he does not believe there is a 10% chance of human extinction, it is better to be overly cautious than to do too little regarding AI safety.
