Microsoft president Brad Smith used a BBC Big Boss Interview podcast released Oct. 12 to declare the White House’s late-September voluntary AI pledge “not sufficient any longer,” and to name four binding rules he wants governments to impose: mandatory incident reports, victim notification, emergency brakes, and pre-release evaluation by independent government bodies.

The framing was deliberately domestic. “Right now, your washing machine is subject to more safety rules than AI,” Smith said. “Every carton of milk that you see in the grocery store is subject to safety rules and inspection.” He added that “AI is moving forward faster than the speed of safety,” and that “That’s not good for the protection of the public. It’s not even good for the industry itself.” His aviation analogy, about a plane flying London to Athens or New York, invoked a system that “has been certified so that it is considered safe by the aircraft regulators in both countries.”

The late-September pledge, signed by OpenAI, Anthropic, Google, Nvidia, Meta, AMD, SpaceX and xAI, committed signatories to independent auditors and to preventing systems from hacking “or access technical systems in unintended ways.” Smith called it “a good step,” then said voluntary regimes only bind “the good companies will do what you want.”

He pointed to London rather than Brussels, whose AI Act regulates by risk tier, and praised the UK AI Security Institute launched under Rishi Sunak in 2023. Trump, who argues “whoever wins AI, wins,” reportedly asked Anthropic and OpenAI last month to withhold new models from British testers; AISI confirmed Monday it still has access. UK AI Minister Kanishka Narayan has told Reuters Britain would consider mandatory rules if voluntary ones fail.

The 2008 TARP sequence applies here: once the biggest players request the rules, the rules usually arrive, and they arrive downstream.

Sources

Sources