AI is less regulated than milk, says Microsoft president
The milk line lands harder than any white paper. We've spent a century figuring out how to keep a perishable liquid safe, pasteurizing it, labeling it, inspecting the dairies, and yet a system that can reason, write code, and influence elections gets a patchwork of voluntary pledges. I think Brad Smith is right to point that out, even if Microsoft has its own reasons for wanting rules it helped draft.
The UK as the leader is an interesting bet. It has the regulatory tradition and the political will, but it's also smaller than the US and China, which are the actual engines of frontier AI. Still, someone has to move first. If London sets a standard that others copy, that's how these things often start.
What I'd want to know next is what those rules actually look like. A global framework could mean transparency requirements, safety testing, or something closer to the FDA model. The comparison to milk suggests we need hard standards, not just principles. I hope the conversation keeps the weirdness of the technology in view. We're not regulating a commodity here. We're deciding how much autonomy we hand to systems we barely understand.