Microsoft's Brad Smith: AI Is Less Regulated Than a Carton of Milk

Microsoft president Brad Smith says artificial intelligence faces fewer safety rules than a carton of milk — and he wants the United Kingdom to close the gap.

"Right now, your washing machine is subject to more safety rules than AI," Smith told ITV in an interview broadcast Friday. "Every carton of milk that you see in the grocery store is subject to safety rules and inspection."

The line is blunt, but the underlying argument is straightforward: everyday consumer goods carry legally mandated safety oversight, while one of the most consequential technologies of the decade operates largely without it. Milk cannot hack companies and governments, Smith pointed out. AI can.

His comments land at a moment when the world's major AI powers are pulling in opposite directions. The United States, under President Donald Trump, is pushing for acceleration. The United Kingdom, in Smith's framing, has an opening to lead on regulation instead.

What did Smith actually say?

Smith made two core claims in the ITV interview.

First, the regulatory gap is real and absurd on its face. Washing machines and grocery-store milk are inspected under safety rules. Frontier AI systems — tools capable, in Smith's words, of putting "the world's cybersecurity at risk" — are not subject to equivalent oversight.

Second, that gap creates a geopolitical opportunity for Britain. Smith argued that the UK occupies a unique position because it is not viewed in Washington the way the European Union is.

"In Washington DC, people don't look at the UK and they don't look at London the way they look at Brussels — that's a strength," Smith said. "I hope that's the role we can play as a company. I hope that's the role the UK can play as a country."

The implication is that Britain can regulate AI credibly without triggering the defensive reflex that EU rules often provoke in the US capital. Smith did not spell out specific legislation he wants the UK to adopt. But he framed the stakes in explicit terms.

"[W]hen you look at models that are so powerful that they're putting the world's cybersecurity at risk, I would say: the moment has arrived," he said.

Why does the US position matter?

Smith's remarks draw force from the contrast with the Trump administration's posture.

Trump has said the US AI industry needs to accelerate — not slow down — to stay ahead of China. Rather than government regulation, Trump prefers that AI companies regulate themselves.

That preference produced a concrete artifact: a self-regulation pact signed at Trump's so-called Super Intelligence summit. The signatories amount to a roll call of American AI power:

• Google CEO Sundar Pichai
• Anthropic CEO Dario Amodei
• Meta CEO Mark Zuckerberg
• OpenAI President Greg Brockman
• Elon Musk
• NVIDIA CEO Jensen Huang

The pact rests on several key tenets, according to the source report. Companies commit to implementing self-regulation, building safeguards to keep pioneer models from going rogue, and using "independent auditors" to assess their models.

What did critics say about the self-regulation pact?

The pact has not quieted skeptics. Critics quoted by the BBC attacked it on two fronts.

One critic said the arrangement lets firms "define what counts as safety" — meaning the companies being regulated decide the standards they are held to.

Another dismissed the agreement outright, telling the BBC that "the accord should be ignored for the distraction it is."

The criticism targets a structural problem rather than a specific clause. Independent auditors only matter, the argument goes, if the audited company sets meaningful thresholds for failure. A voluntary pact, signed voluntarily, enforced voluntarily, leaves that decision in private hands.

This is the gap Smith's milk analogy is designed to expose. No carton of milk gets to choose its own inspection regime. No washing machine manufacturer writes the safety standard its product is tested against. AI labs, under the current US approach, effectively do.

Is the industry itself split?

Yes — and that is what makes Smith's intervention notable.

Major US AI companies, including Anthropic, OpenAI and Google's DeepMind, have spoken in favor of an AI development slowdown. That is a striking position for commercial labs racing to build ever-larger systems. It suggests at least parts of the industry see risks they believe current safeguards cannot contain.

Yet the White House is moving in the opposite direction, framing AI dominance as a strategic competition with China that the US cannot afford to lose.

Microsoft sits in an unusual position within that split. It is one of the largest investors in the AI supply chain, and its president is now publicly arguing that the sector his company helps fund is under-regulated relative to dairy products. Smith is not calling for a slowdown by name in the interview — he is calling for rules, and specifically for Britain to write them.

Why the UK and not the EU?

Smith's geography argument deserves attention on its own terms.

Brussels, home of the European Union's regulatory apparatus, is watched warily in Washington. EU rules on technology have repeatedly collided with American tech interests, and the mention of Brussels in US policy circles often signals friction.

London, in Smith's telling, does not carry that baggage. "People don't look at the UK and they don't look at London the way they look at Brussels — that's a strength," he said.

The strength is twofold. Britain has the technical and institutional weight to regulate AI seriously — a legacy financial-regulation infrastructure, a working legal system, and a credible standards tradition. And it lacks the adversarial history with US tech that makes EU regulation a lightning rod in Washington.

Smith made clear he sees Microsoft as a partner in that project, not merely a commentator: "I hope that's the role we can play as a company."

Whether UK policymakers accept the framing is a separate question. A regulatory invitation from the president of one of the world's largest AI companies cuts both ways. It signals industry tolerance for rules — or at least a preference for London's rules over Brussels's. It also raises the obvious concern that incumbent players may shape whatever regime emerges.

What happens next?

The immediate pressure point is the divergence between the two approaches now in play.

In the US, the Super Intelligence pact represents the self-regulation path: voluntary commitments, industry-defined safety, independent auditors chosen under terms the companies help set. Its critics already consider it hollow.

In the UK, Smith has laid down a marker that the moment for binding rules has arrived, and that Britain is the place to write them — precisely because Washington distrusts Brussels but not London.

The open question is whether the UK government moves to claim that role, or whether the absence of US federal action persists while voluntary commitments remain the strongest constraint on frontier AI development. Smith's own words frame the deadline: when models are powerful enough to threaten the world's cybersecurity, waiting is no longer a neutral choice.