AI faces less regulation than milk, Microsoft president warns

Microsoft’s Brad Smith says AI has fewer safety rules than milk or washing machines, and sees a UK opening as US AI firms sign a self-regulation pact — with practical stakes for content teams.

What happened

Microsoft President Brad Smith told ITV on Friday that AI is less regulated than everyday consumer goods. "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." Engadget reports that milk can't hack companies and governments like AI definitely can, but it is still subject to more regulation.

Major US AI giants like Anthropic, OpenAI and Google's DeepMind have spoken in favor of an AI development slowdown. However, US President Donald Trump has said that, "if anything, the US AI industry needs to accelerate to stay ahead of China." Rather than government action, Trump would prefer that AI companies regulate themselves.

At Trump's so-called Super Intelligence summit, AI bosses including Sundar Pichai, Anthropic's Dario Amodei, Mark Zuckerberg, OpenAI President Greg Brockman, Elon Musk and NVIDIA CEO Jensen Huang signed a pact. It includes self-regulation, safeguards to keep pioneer models from going rogue and using "independent auditors" to assess models. Critics were unimpressed, with one telling the BBC it allows firms to "define what counts as safety" and another saying "the accord should be ignored for the distraction it is."

Key facts

• Microsoft President Brad Smith said on ITV: "Right now, your washing machine is subject to more safety rules than AI," and every carton of milk is subject to safety rules and inspection.
• Anthropic, OpenAI and Google's DeepMind have supported an AI development slowdown, while Trump said the US AI industry needs to accelerate to stay ahead of China.
• A Super Intelligence summit pact was signed by Sundar Pichai, Anthropic's Dario Amodei, Mark Zuckerberg, OpenAI President Greg Brockman, Elon Musk and NVIDIA CEO Jensen Huang, covering self-regulation, safeguards for pioneer models and independent auditors.
• Critics told the BBC the pact lets firms "define what counts as safety" and should be ignored as a distraction.
• Smith said Washington DC doesn't view the UK and London the way it views Brussels, giving the UK a unique opening and saying "the moment has arrived" for powerful models that put cybersecurity at risk.

Our analysis

The core tension for operators is not political; it is definitional. If AI safety remains voluntary, creators and brand teams will likely see faster release of generative tools, but the companies building those tools may also define what "safety" means on their own terms. One BBC critic's wording matters because it suggests vendors could label a model as audited without a shared standard for what independent review must include.

Smith's framing also suggests a likely regulatory patchwork. His statement that Washington does not look at the UK the way it looks at Brussels implies the UK could become a more approachable venue for AI rules than the EU. For global social and content teams, that may create uneven duties: an AI feature available in one market could face different disclosure, training-data or security expectations in another.

The mention of "world's cybersecurity at risk" is the strongest signal for operators. It suggests that AI risk is moving from abstract ethics to operational trust, which affects automated content pipelines, synthetic media detection and vendor due diligence.

What it means for operators

• Ask AI vendors what "independent auditors" means in practice: who audits, what is checked, and whether reports are public. Do not treat summit pacts as evidence of compliance.
• Track UK announcements separately from EU and US rules; a UK-led push could affect campaigns targeting British audiences before other markets, so build regional flexibility into AI content workflows.
• Document how AI is used for captions, images, video and community responses now. Voluntary self-regulation periods often become mandatory disclosure later, and retroactive cleanup is costly.
• For crypto communities and creators using AI agents, add security review to model adoption: if pioneer models can put cybersecurity at risk, account takeover and synthetic impersonation are practical trust issues, not just policy debates.