Brad Smith Says AI Is Less Regulated than Milk; UK Announces It Will Legislate Oversight
Microsoft president Brad Smith says AI is less regulated than milk and calls for global safety rules including kill switches — on the same day the UK AI minister announced the government will legislate oversight of AI, putting both at odds with Washington's stance.
On October 12, two speeches landed within hours of each other that pointed toward the same conclusion from very different directions. Microsoft president Brad Smith told the BBC that AI faces fewer safety rules than milk and washing machines, and UK AI minister Kanishka Narayan stood up at Imperial College London to announce that the UK government will legislate to increase oversight of artificial intelligence. Together, they outline a regulatory moment that is moving faster than the political consensus Smith’s own employer operates under in Washington.
Smith’s remarks, delivered on the BBC’s Big Boss Interview podcast, were unusually specific for a sitting tech executive calling for his own industry to be hemmed in. He argued for global safety rules that would include kill switches for AI systems, legal obligations to the victims of rogue AI agents, and independent government evaluators tasked with vetting AI products before release — a model closer to pharmaceutical or aviation oversight than anything the tech sector has accepted to date. The exact comparison he reached for was milk and washing machines, both more tightly regulated than AI at the consumer level, which is the kind of contrast that either sounds absurd or clarifying depending on how seriously you take the gap between where AI capability is and where the legal frameworks around it sit.
What made the remarks more striking was where Smith pointed for leadership. Not the EU, not the United States, but the UK — and the reason he gave was institutional, not sentimental. Smith argued the UK is best positioned to lead on AI regulation because of the role its AI Security Institute already plays in evaluating frontier models, giving London something neither Washington nor Brussels has in its current form: a standing government body with direct, hands-on access to the systems it’s being asked to govern. That the AI Safety Institute exists at all, and gives the UK a regulatory node other democracies haven’t built yet, is what makes Smith’s endorsement more than a diplomatic compliment.
It was also a pointed divergence from the direction of travel in his own country. Smith’s comments place him directly at odds with the Trump administration’s preference for industry self-regulation on AI, a stance that has already surfaced in concrete form: Trump reportedly asked Anthropic and OpenAI to withhold new models from British testers, though the AI Safety Institute confirmed it can still access and test the latest frontier models regardless. Smith made a specific argument about why the UK can regulate in ways Washington cannot right now — that the US doesn’t view London the same way it views Brussels, meaning UK oversight faces fewer of the geopolitical headwinds that complicate EU regulation — which is as close to a formal break with his own government’s position as a Microsoft president is likely to get.
The UK side of the equation was already moving on its own. Speaking the same day at Imperial College London, [Kanishka Narayan warned that it is