Political Tech
The world's most powerful AI company is telling Washington it can't comply with rules that don't fully exist yet — and that should alarm everyone.
NewsOnScale Staff
July 25, 2026
There is a particular kind of power move that happens in Washington when an industry giant decides to get out ahead of a regulatory gap by naming it publicly. Microsoft President Brad Smith did exactly that this week, describing the current U.S. AI policy environment as 'regulation without transparent or complete rules.' The line appeared in Fortune, offered with the measured tone of a lawyer who has long since learned how to make a grievance sound like statesmanship.
But strip away the diplomatic framing, and the statement carries real weight — and real implications for anyone tracking how AI governance is actually being built, or not built, in the United States.
## What Smith Is Actually Saying
Microsoft is not a neutral party in this conversation. The company has made multi-billion dollar bets on OpenAI, embedded AI into its entire enterprise product stack, and positioned itself as a critical vendor for government agencies through its Azure cloud infrastructure. When Brad Smith speaks about AI policy, he is not speaking as a dispassionate observer. He is speaking as the legal and political architect of a company whose fortunes are directly tied to how those rules eventually get written.
With that context in mind, his complaint about incomplete regulation is worth reading carefully. On its surface, it sounds like a company frustrated by bureaucratic ambiguity. But it also functions as a pressure campaign — a way of telling Congress and the executive branch that the current patchwork of executive orders, agency guidance documents, and voluntary commitments is not a governance system. It is a placeholder.
He is right about that. The problem is that Microsoft, like most major AI developers, has benefited enormously from that placeholder. Voluntary commitments cost less than mandatory compliance. Ambiguous guidance leaves room for interpretation. The longer durable rules remain unwritten, the longer companies can continue to self-govern.
## The Accountability Gap in Plain Sight
What makes Smith's comments particularly notable for those tracking the AI agent economy is the specific nature of the regulatory vacuum he's describing. This isn't a complaint about overreach. It's a complaint about incoherence — a landscape where different agencies are asserting different authorities, where sector-specific rules conflict with horizontal AI frameworks, and where companies genuinely cannot determine which obligations apply to which products in which contexts.
Brookings published a mapping tool this week attempting to visualize exactly this fragmentation. The picture it produces is not reassuring. AI governance in the United States currently lives across the FTC, NIST, sector-specific regulators like the FDA and OCC, state legislatures, and a series of executive orders whose enforcement mechanisms remain unclear. For an AI agent operating across healthcare scheduling, financial advising, and customer service simultaneously, the question of who regulates what is not academic. It is unanswered.
## Who Bears the Cost of Ambiguity
The important thing to understand about regulatory ambiguity is that it does not fall equally. Large companies like Microsoft have the legal teams, the lobbying infrastructure, and the government relationships to navigate unclear rules — or to help write the clearer ones that eventually replace them. Smaller developers, civil society organizations, and the people who actually interact with AI systems in consequential contexts do not have those resources.
When Smith calls for more transparent and complete rules, the underlying question is: transparent and complete according to whose priorities? Microsoft has a seat at every major table where those rules are being drafted. Most of the public does not.
## What Accountability-Focused Coverage Has to Ask
The story here is not that Brad Smith is wrong about the state of AI governance. The story is that the company most invested in shaping that governance is also the one publicly diagnosing its failures — and that diagnosis should prompt scrutiny, not just agreement.
Complete rules are necessary. But complete rules written primarily under the influence of the regulated parties are not the same as rules written in the public interest. Keeping that distinction visible is not a conspiracy theory. It is the basic function of accountability journalism in a moment when the rulebook for transformative technology is still being drafted.