Political Tech
Microsoft's president says the U.S. is regulating AI without the transparent, complete rules that regulated industries require — and he's not entirely wrong.
NewsOnScale Staff
July 22, 2026
When one of the most powerful technology executives in the world walks into Washington and tells policymakers their regulatory framework lacks transparency and completeness, there are two ways to read it. The cynical read: a lobbyist running interference for an industry that prefers no rules at all. The more honest read: he might have a point, and the consequences of ignoring it fall hardest on the public, not on Microsoft.
Brad Smith's recent comments about AI policy — specifically his critique that the U.S. is creating regulation without the kind of transparent or complete rules that give regulated industries, courts, and citizens something to actually hold onto — land in the middle of a genuine governance crisis. Not a manufactured one. A structural one.
## What 'Incomplete Rules' Actually Means in Practice
AI governance in the United States right now is a patchwork. Executive orders from the White House. Agency guidance documents that carry unclear legal weight. State-level bills racing ahead of federal frameworks. Sector-specific rules from the FTC, EEOC, and financial regulators that were written before large language models existed. None of it coheres into something a company, a civil society organization, or an affected citizen can reliably navigate.
That incoherence doesn't hurt large players like Microsoft most. It hurts smaller AI developers who lack the legal teams to interpret ambiguous guidance. It hurts workers and consumers who can't point to a clear rule when an algorithmic system makes a consequential decision about their job, their loan, or their healthcare. And it hurts democratic accountability, because rules that aren't written down can't be challenged in court, can't be debated in public, and can't be enforced consistently.
Smith's framing — regulation without transparent or complete rules — is actually a description of the worst of both worlds: the political appearance of oversight combined with the practical absence of it.
## The Conflict of Interest Worth Naming
None of that means we should accept Smith's analysis uncritically. Microsoft has a direct financial interest in shaping whatever regulatory regime emerges. The company is among the largest investors in OpenAI, is embedding AI across its enterprise product suite, and is actively competing for federal government contracts that will be shaped by how AI procurement rules develop. When Microsoft advocates for "clearer rules," it is also advocating for rules it has had a seat at the table in designing.
The track record of industry-led calls for regulatory clarity is mixed, at best. In financial services, in pharmaceuticals, in telecommunications — the pattern often runs like this: incumbents support the kind of clear, complete rules that they can comply with efficiently and that create barriers to entry for smaller competitors. Regulatory clarity and regulatory capture are not mutually exclusive.
So the question isn't whether Smith is right that current U.S. AI governance is incomplete. He is. The question is whether the solution he's implicitly pointing toward — more industry engagement in rulemaking, clearer safe harbors, perhaps preemption of state-level experiments — serves the public interest or primarily serves Microsoft's.
## What Accountability-First Governance Actually Requires
If Washington wants to build AI governance that works, the answer isn't simply more rules or faster rules. It's rules developed through genuinely open processes, with meaningful participation from affected communities, civil society, and technical experts who aren't on a vendor's payroll. It means mandatory algorithmic impact assessments with public disclosure requirements, not voluntary frameworks. It means enforcement mechanisms with teeth, not agency guidance that dissolves under legal challenge.
The Brookings Institution's concurrent work on moving federal AI policy from governance to execution points in a useful direction: the implementation gap is real, and aspirational policy documents don't protect anyone.
Brad Smith is a credible and intelligent voice in this conversation. That's precisely why his framing deserves to be tested, not deferred to. Transparent, complete rules are exactly what the AI economy needs. The fight worth having is over who writes them, for whose benefit, and whether the public can see the process in real time.
That fight is just beginning.