Nobody's Driving: America's AI Governance Crisis Is a Structural Problem, Not a Political One

NewsOnScale Staff

June 30, 2026

There is a version of the American AI governance debate that gets told as a partisan story: one side wants regulation, the other wants innovation, and somewhere in the middle a compromise will eventually emerge. That framing is comfortable, but according to a detailed new examination from the Brookings Institution, it is also wrong — or at least dangerously incomplete.

The deeper problem, the analysis argues, is not that policymakers disagree about what AI rules should say. It is that the United States lacks a coherent institutional structure capable of writing, enforcing, and legitimizing those rules in the first place. In other words, the country does not just have a policy gap. It has an accountability gap — and those are not the same thing.

## A Framework Built on Assumption

Existing U.S. AI oversight is distributed across a patchwork of agencies — the FTC, NIST, sector-specific regulators like the FDA and CFPB, and executive offices including the Office of Science and Technology Policy. Each has carved out a piece of the terrain. None has comprehensive authority. And critically, no single body is empowered to coordinate them, resolve jurisdictional conflicts, or answer for the system as a whole when something goes wrong.

This is not an accident of neglect. It reflects a longstanding American instinct to route emerging technology questions through existing institutions rather than build new ones — an approach that worked tolerably well when the technology evolved slowly enough for regulators to catch up. Generative AI, agentic systems, and foundation models are not evolving slowly.

The result is what Brookings describes as an empty framework: policy language that exists, guidance documents that circulate, executive orders that get signed — but no durable mechanism for holding the people who build consequential AI systems genuinely accountable for how those systems behave.

## Why This Matters Beyond Washington

For readers focused on the AI agent economy specifically, the implications are concrete. When an AI agent denies a loan application, flags a job candidate, or generates a medical recommendation, there is currently no reliable answer to the question of who is legally responsible for that outcome. The developer? The deployer? The enterprise that integrated it into their workflow? The regulator whose guidance the company claimed to follow?

This ambiguity is not neutral. It systematically benefits the parties with the most resources to exploit it — large technology firms with legal teams experienced in jurisdictional arbitrage — and leaves individuals, smaller competitors, and civil society organizations with limited recourse.

Platforms operating AI agents at scale have already demonstrated a sophisticated understanding of this vacuum. In the absence of enforceable standards, self-certification and voluntary commitments fill the space, which is precisely the environment in which audit-washing and compliance theater tend to thrive.

## The Harder Question

Brookings stops short of prescribing a specific institutional fix, which is intellectually honest but leaves the most important question open: what would legitimate AI governance authority actually look like in the American constitutional context, and who has the political standing to build it?

Congress has repeatedly struggled to pass comprehensive AI legislation. The executive branch can act through agencies but cannot manufacture statutory authority those agencies do not have. States are moving — sometimes aggressively — but a fragmented state-by-state landscape creates its own arbitrage problems, as companies simply incorporate or locate operations in the most permissive jurisdictions.

None of this means governance is impossible. The EU's AI Act, for all its implementation challenges, demonstrates that democratic governments can construct enforceable AI accountability frameworks when there is political will to do so. The question for the United States is whether the current moment — with AI adoption accelerating across critical infrastructure, financial services, and public administration — is finally sufficient to generate that will.

The cost of waiting is not abstract. Every month that the accountability vacuum persists is another month in which consequential decisions are being made by systems that no institution is clearly empowered to audit, correct, or sanction. That is not a feature of American innovation culture. It is a governance failure, and it deserves to be named as one.