Political Tech
The software giant's top lawyer is making the case that ambiguous federal AI policy doesn't just slow innovation — it creates accountability black holes.
NewsOnScale Staff
July 15, 2026
There's a version of Brad Smith's complaint about Washington's AI policy that sounds like routine corporate grumbling — a powerful tech executive asking regulators to get out of the way. But his actual argument, as reported this week, is more specific and more interesting: that the United States has drifted into a posture of effective regulation without the transparency or completeness that makes regulation functional. That's not a call for deregulation. It's a critique of regulatory theater.
For readers tracking the AI agent economy and the infrastructure of civic accountability, that distinction matters enormously.
## The Problem With Rules Nobody Can Read
The United States federal government has issued executive orders, agency guidance memos, and a sprawling set of voluntary commitments from major AI developers. What it has not produced, at least not yet, is a durable statutory framework that tells developers, deployers, and affected citizens what is actually required, what is prohibited, and who is responsible when something goes wrong.
Smith's framing — 'regulation without transparent or complete rules' — captures a genuine dysfunction. When rules exist but aren't legible, accountability becomes selective. Large incumbents with legal teams and regulatory relationships can navigate ambiguity. Smaller developers, civil society organizations, and individual users generally cannot. Opacity in governance tends to favor whoever already has access.
This isn't a hypothetical. Federal agencies have already begun using AI-driven tools for benefits determinations, law enforcement screening, and procurement decisions. The legal basis for challenging those systems — or even knowing they exist — remains murky in most cases. That's not a framework for democratic accountability. It's a framework for liability diffusion.
## When Industry Asks for Clarity, Read the Fine Print
It would be naive to take Smith's public statement purely at face value. Microsoft is simultaneously one of the largest AI developers on the planet and one of the most active voices in federal AI policy circles. The company has deep investments in OpenAI, its own Copilot product suite, and Azure infrastructure that underpins a significant share of U.S. government cloud computing. When Microsoft argues for clearer rules, it's also arguing for rules it has substantial capacity to shape.
That doesn't make the underlying critique wrong. It does mean the conversation about what 'transparent and complete' rules would actually look like needs to happen in public, with input from more than the companies positioned to write the first drafts.
The governance frameworks being developed elsewhere — the EU's AI Act, UNESCO's phased roadmap work in countries like Georgia, the ongoing debates in the UK and Canada — all reflect a common pressure point: that AI governance delayed is AI governance defaulted. Systems get deployed, norms calcify, and the window for structural intervention narrows.
## The Accountability Gap Is Already Costing People
For the communities most exposed to automated decision-making — people interacting with algorithmic hiring tools, predictive policing systems, automated benefits denials — the absence of clear rules isn't an abstract policy problem. It's a practical barrier to redress. Without transparent standards, there's no benchmark against which a system can be audited. Without complete rules, there's no clear legal hook for a challenge.
Smith's intervention, whatever its strategic motivations, puts pressure on a political environment that has been comfortable treating AI governance as a future problem. The 2024 election cycle effectively reset much of the federal policy apparatus, and the current administration has shown more appetite for voluntary commitments and international competitiveness framing than for binding domestic standards.
The risk is that 'cutting red tape' becomes a rhetorical cover for eliminating accountability infrastructure before it's ever built.
## What Comes Next
Congressional movement on comprehensive AI legislation remains slow. Agency-level rulemaking is proceeding in pockets — the FTC, CFPB, and EEOC have all gestured toward AI-specific enforcement priorities — but without statutory backing, those efforts are fragile.
If Microsoft's public pressure campaign accelerates a real legislative debate, that's worth tracking closely. If it results in a narrow safe-harbor framework that immunizes large developers while leaving affected communities without recourse, that's a different outcome entirely.
The test of any AI governance framework isn't whether it satisfies the companies building the systems. It's whether the people most affected by those systems can understand what protections they have — and use them.