Microsoft’s Brad Smith urges binding AI safety rules

Microsoft President Brad Smith is calling for legally binding safeguards for artificial intelligence, including emergency shutdown mechanisms, incident reporting and independent assessments of advanced systems. He also argues that the United Kingdom is well placed to help shape international safety standards, but the proposals have not been adopted as law. In interviews published over the weekend and on Monday, Smith said AI faces fewer safety requirements than everyday consumer products such as milk and washing machines. He called for companies to be legally required to report incidents involving rogue AI systems and notify people whose systems are affected. He also proposed emergency brakes that could pause or slow an AI system when it behaves unexpectedly. Smith said independent evaluators, including government bodies, should assess new AI technology before release. He has described voluntary industry commitments to independent audits as a necessary step but not a sufficient substitute for legal requirements. His proposals address oversight of systems that can act in ways their operators did not intend.