In an era when artificial intelligence is reshaping nearly every dimension of human life, Microsoft's president has surfaced a disquieting truth: the systems most capable of altering society at scale are governed by fewer formal safeguards than a carton of milk. The company is calling not for voluntary pledges but for binding international frameworks, with the United Kingdom identified as a potential anchor for global standards. It is a rare moment when a technology giant advocates for the very constraints that might slow its own momentum — a signal that those closest to the technology sense t
Microsoft President Warns AI Faces Fewer Safety Rules Than Milk, Calls for Global Regulation
AI faces fewer safety rules than milk
Mark
So when Microsoft says AI is less regulated than milk, what exactly are they pointing to? What rules does milk have that AI doesn't?
Mimi
Milk has to be tested for pathogens, labeled with ingredients and allergens, kept at specific temperatures, traced back to its source if something goes wrong. AI has almost none of that. No mandatory safety testing before deployment, no labeling of training data, no traceability.
Luke
But can you actually compare them? Milk safety is about preventing immediate physical harm. AI harms are often indirect—bias, manipulation, misinformation. The regulatory models aren't equivalent.
Mimi
That's fair. But the point isn't that the rules should be identical. It's that we have *some* framework for milk and almost nothing for AI, even though AI affects billions of people.
Mark
Why is Microsoft pushing for this? Don't regulations slow them down?
Mimi
Yes, they do. But Microsoft's leadership seems to believe the risks of an unregulated AI landscape are worse than the friction of compliance. If AI causes a major failure—a system that discriminates at scale, or spreads harmful misinformation—the backlash could be much worse than regulation would be.
Luke
That's a business calculation, though. We should be clear about that. It's not pure altruism.
Mimi
No, it's not. But it's also not cynical. They're saying: we need rules, and we need them globally, because no single company can police itself well enough.
Mark
Why the UK specifically?
Mimi
They've been developing AI governance approaches and could set a standard that other countries might follow. If a major economy establishes credible rules, it creates pressure on others to do the same.
Luke
The hard part is still ahead—actually writing those rules. How do you regulate something that changes every six months? How do you detect harm that's subtle and distributed?
Mimi
Those are the real questions. The milk comparison gets attention, but solving it is much harder than the comparison suggests.
Le Pouls
- AI systems are being deployed at massive scale with no mandatory safety testing, no required disclosures, and no regulatory body empowered to approve or reject them before they reach the public.
- Microsoft's leadership has made the gap viscerally clear: milk travels through a century of mandated oversight while AI, capable of affecting millions simultaneously, operates in a largely ungoverned space.
- The call for coordinated global regulation — not industry self-policing — is striking precisely because it comes from a company with enormous financial incentive to move fast and avoid friction.
- The United Kingdom has been positioned as a potential leader in establishing credible AI governance standards that other economies might adopt, creating a baseline floor for the industry.
- The hardest problem remains unsolved: unlike food contamination, AI-driven harms can be invisible, slow-moving, and nearly impossible to trace back to a single point of failure or accountability.
In an era when artificial intelligence is reshaping nearly every dimension of human life, Microsoft's president has surfaced a disquieting truth: the systems most capable of altering society at scale are governed by fewer formal safeguards than a carton of milk. The company is calling not for voluntary pledges but for binding international frameworks, with the United Kingdom identified as a potential anchor for global standards. It is a rare moment when a technology giant advocates for the very constraints that might slow its own momentum — a signal that those closest to the technology sense the stakes of inaction most acutely.
Microsoft's president has offered a deliberately jarring comparison to illustrate a growing crisis of governance: artificial intelligence, one of the most consequential technologies ever developed, is subject to fewer safety requirements than milk. The contrast is meant to be uncomfortable — and it is.
Milk moves through a heavily regulated supply chain, with mandatory testing, labeling, and traceability at every stage. AI systems, by contrast, are deployed at scale with no government approval process, no standardized safety protocols, and no required disclosure of how they were built or what harms they might cause. Policy has simply not kept pace with the technology.
What makes Microsoft's position notable is the source. This is a company with vast financial incentive to move quickly and avoid regulatory friction. That its leadership is calling for binding legal frameworks — not voluntary guidelines — suggests the risks of inaction are being weighed seriously even by those who stand to be constrained. The company has also pointed to the United Kingdom as a nation well-positioned to lead, with the hope that credible standards established by one major economy could set a floor others would follow.
Yet the deeper challenge remains largely unaddressed. AI safety does not map neatly onto food safety. A contaminated product causes traceable harm within days; an AI system trained on flawed data may quietly discriminate against millions over years before anyone notices. The mechanisms for detecting harm, assigning responsibility, and enforcing compliance are still more theoretical than real.
Microsoft's warning carries an implicit admission: no single corporation can establish the universal standards the public actually needs. That requires governments, political will, and international cooperation — a willingness to constrain a technology that many nations see as central to their economic futures. We have known how to regulate food for a century. For AI, we are still improvising.
Microsoft's president has drawn a stark comparison to make a point about how little oversight artificial intelligence actually receives. The company's leadership argues that AI systems face fewer safety requirements than milk does—a deliberately jarring contrast meant to underscore what they see as a dangerous regulatory vacuum.
The concern centers on a fundamental asymmetry. Milk, a commodity produced by the millions of gallons annually, moves through a heavily regulated supply chain. Testing, labeling, temperature control, traceability—all of it is mandated and monitored. AI systems, by contrast, are deployed at scale with minimal formal safety requirements. No government agency must approve an AI model before it reaches users. No standardized testing protocol exists. No mandatory disclosure of training data or potential harms is required. The technology advances faster than policy can follow.
Microsoft's position reflects a broader anxiety within the technology sector itself. The company is calling for coordinated global regulation—not voluntary guidelines, not industry self-policing, but actual legal frameworks that would establish baseline safety standards across borders. This is notable because it comes from a company with enormous financial incentive to move fast and avoid friction. That Microsoft is advocating for the very constraints that might slow its own work suggests the leadership believes the risks of inaction outweigh the costs of regulation.
The company has also identified the United Kingdom as potentially the best positioned nation to lead this effort. Britain has been developing AI governance approaches and could serve as a model for other countries attempting to balance innovation with safety. The idea is that if one major economy establishes credible standards, others may follow, creating a floor below which AI development cannot fall.
What remains unresolved is how such regulation would actually work in practice. AI safety is not like food safety—the risks are not always visible, the failures not always immediate or traceable. A contaminated batch of milk sickens people within days. An AI system trained on biased data might discriminate against millions of people over years without anyone noticing. The mechanisms for detecting harm, assigning responsibility, and enforcing compliance remain largely theoretical.
Microsoft's warning also carries an implicit acknowledgment: the company cannot solve this alone. No single corporation, no matter how large or well-intentioned, can establish the kind of universal standards that would actually protect the public. That requires governments to act, which means it requires political will, international cooperation, and the willingness to constrain a technology that many see as central to economic competitiveness. The milk comparison is memorable precisely because it highlights how far behind we are—we have known how to regulate food for a century, but we are still improvising rules for technology that is reshaping the world in real time.
Citations marquantes
AI currently operates with fewer safety requirements than milk, despite affecting billions of people globally— Microsoft leadership position
Envie de l'histoire complète ?
Lire l'original sur Google News ↗