The company building Copilot just told governments that AI companies have less oversight than dairy farmers.
The Summary
- Microsoft President Brad Smith says milk and washing machines face stricter safety rules than AI systems, calling for global regulatory standards
- Smith positioned the UK as the ideal leader for AI regulation, citing its tech sector and regulatory experience
- He's advocating for mandatory incident reporting when AI systems cause harm, similar to aviation safety protocols
The Signal
Brad Smith delivered a blunt assessment: the AI industry operates with less regulatory oversight than milk producers or washing machine manufacturers. This isn't abstract concern. Smith is saying the systems his company ships to 400 million Windows users have fewer safety requirements than the appliance in your laundry room.
The Microsoft president's timing matters. After the EU AI Act passed and several US states proposed their own frameworks, the regulatory landscape is fragmenting. Smith wants the UK to step in as coordinator. His reasoning: Britain has both a significant AI industry and institutional experience with tech regulation through bodies like the ICO and CMA.
"AI companies face less regulatory scrutiny than dairy farmers or appliance makers."
Smith's specific proposal centers on mandatory incident reporting. When an AI system causes measurable harm, companies would be required to document and report it to regulators. The parallel he drew was aviation, where every significant incident feeds into safety databases that improve the entire industry. Right now, when an AI hiring tool discriminates or a content moderation system fails, there's no systematic way to track patterns or force corrections.
The unusual part: Microsoft is asking for regulation. Not fighting it, not lobbying against it, actively requesting it. This tracks with their recent pattern:
- Supported the EU AI Act while Meta and others pushed back
- Backed California's AI safety bills that OpenAI opposed
- Published their own Responsible AI Standard before being required to
Why this specific ask matters:
Microsoft runs Azure, which hosts a large percentage of AI development globally. They know what's being built. They see the liability exposure. They're positioned to absorb compliance costs that would crush smaller competitors. Calling for regulation when you're the incumbent with deep pockets is smart defensive strategy.
But there's genuine operational logic too. Microsoft sells AI tools to enterprises in healthcare, finance, and government. Those customers want clear compliance frameworks. They want to know what "safe deployment" means in legal terms. The current ambiguity slows enterprise adoption, which slows Microsoft's revenue growth.
The UK angle is strategic maneuvering. Post-Brexit Britain is looking for global leadership opportunities. Its AI Safety Institute is already doing technical evaluations. Smith is offering a path: convene global standards, build the reporting infrastructure, become the trusted third party that validates AI systems for international markets.
The Implication
Watch for the UK government's response in the next quarter. If they take Smith's pitch seriously, expect them to propose an international AI incident database and coordinate testing standards with the EU and US. That would give British institutions real influence over what gets deployed globally.
For companies building AI agents or automation tools: assume mandatory incident reporting is coming within 18 months in major markets. Start documenting failures and safety testing now. The aviation model Smith referenced took decades to build. AI won't get decades. You'll need those systems before regulators ask.