Microsoft president calls for global AI safety rules

Summary

Microsoft president Brad Smith says artificial intelligence is less regulated than products such as milk and washing machines, and argues that global safety rules are now urgent. He proposes legal requirements for companies to report serious AI incidents and notify people affected when a system is breached or goes rogue. Smith also wants AI systems to include emergency brakes, sometimes called kill switches, so they can be paused or slowed when they become unsafe. He supports independent government evaluators signing off on some AI products before release, drawing on safety practices used for other products. Smith said the UK is well placed to lead international standards because of its technical capability and AI Security Institute, and suggested Washington may be more receptive to UK-led rules than standards associated with the EU. His position differs from President Donald Trump’s preference for industry self-regulation to preserve US competitiveness against China. Smith welcomed a recent White House-backed commitment by major technology companies to work with independent auditors, but said voluntary self-regulation is insufficient because responsible firms are not the only companies that need oversight.