The AI moratorium in H.R. 1 isn’t a policy compromise. It’s a power play.
In a single stroke of the pen, the federal government seeks to block states and cities from creating just about any AI-specific rules—no matter how targeted, protective, or future-facing. And the federal government steps back by offering no federal replacement and no new oversight body.
Just silence. And that’s the strategy.
This isn’t about building a national framework. It’s about setting a precedent: scale fast, self-regulate, and let the market write the rules.
What the Bill Says
Under § 43201(c)(1), H.R. 1 blocks:
“…any law or regulation… limiting, restricting, or otherwise regulating artificial-intelligence models, artificial-intelligence systems, or automated-decision systems entered into interstate commerce.” 1
That covers a lot of ground: hiring algorithms, credit scoring, recommender systems—anything with a computational output that nudges a decision.
Two carve-outs exist under § 43201(c)(2):
- Facilitation laws: States can streamline AI deployment (think zoning, incentives)
- Neutral laws: General-purpose laws that treat AI and non-AI the same
State criminal statutes are also left alone. So if AI’s used to commit fraud or harassment, enforcement can still happen. But proactive governance, risk audits, and accountability rules are expressly blocked under this bill.
No New Guardrails, Just Fewer Actors at the Table
The bill doesn’t create a national oversight regime. It doesn’t task agencies with enforcement. It just clears the field and lets the federal government decide—later—if it wants to show up.
This isn’t a framework. It’s a pause.
For a decade, AI policy gets centralized—but not necessarily exercised.
This Is Regulatory Displacement, Not Delay
What looks like a jurisdictional reshuffle is actually a deregulatory playbook. Let’s be explicit:
- States are blocked from acting for 10 years.
- No new federal structure fills that vacuum.
- No obligation exists for federal agencies to intervene.
- Only laws that help deployment or treat AI like ordinary software are allowed.
This isn’t buying time to regulate later. It’s buying room to entrench norms now, before regulation becomes possible.
The Pattern: Preempt, Clear, Entrench
This move doesn’t stand alone. It fits a pattern:
- The U.S. Copyright Office shakeup removed key friction for AI model training—dislodging a regulatory body that had become increasingly vocal about licensing and consent.
- In Bartz v. Anthropic, a federal court ruled that using copyrighted materials to train AI systems can qualify as fair use. It extended an old doctrine into a new domain—and in doing so, cleared a major legal hurdle for AI firms.
- Now, H.R. 1 eliminates the next layer of constraint: local attempts to introduce friction through transparency rules, liability standards, or data rights.
None of these steps require collusion. The system doesn’t need a backroom strategy. It works through alignment.
This isn’t conspiracy—it’s convenience. The danger isn’t that agencies coordinated. It’s that they didn’t need to. In a fragmented system, aligned inertia can function as strategy. That’s how power scales: not through command, but through convergence.
When each branch—executive, judicial, legislative—acts independently but in directionally consistent ways, the result is functionally indistinguishable from design.
The vacuum isn’t accidental. Industry norms are the intended filler.
The outcome isn’t oversight later; it’s irrelevance of oversight later.
Once deployment scales, AI companies become regulators in everything but name—not because they write laws, but because their systems become the infrastructure government and industry rely on to function.
We’ve seen this before:
- In healthcare, private insurers set the defaults that public systems follow.
- In payments, processors gate access to the economy more directly than regulators.
- In defense, private contractors own the stack government operates on.
Once public governance relies on private infrastructure, the ability to regulate becomes a negotiation—not a mandate.
Not every firm wants deregulation; some will want clarity to outlast chaos. But H.R. 1 doesn’t reward stability—it rewards incumbency. The absence of early constraint advantages those already deployed. Regulation delayed becomes competition denied.
With H.R. 1, AI is headed a similar direction: entrench first, defer oversight, and let the market make policy by making itself indispensable.
What This Forces You to Accept
| Assumption | Strategic Consequence |
|---|---|
| AI deployment is inherently good | Oversight is positioned as threat, not part infrastructure |
| Experimentation advances the cause | Even if that experimentation means a uniform lack of accountability |
| States cannot be trusted to govern emerging tech | Even when they’ve led in environmental, consumer, or data regulation |
| Private-sector momentum should take precedence | Especially if public friction slows adoption or reduces competitiveness |
This isn’t federal oversight delayed. It’s public recourse denied.
What The Bill Doesn’t Do
This part’s simple, and maybe the most important:
- No national rules
- No federal regulator
- No required transparency or risk disclosures
- No distinction between low-risk consumer apps and high-stakes AI in employment, finance, or criminal justice
- No guarantee of future action
The only certainty is that states are benched for ten years. After that, the moratorium expires—unless Congress chooses to extend it.
This Isn’t a Timeout. It’s a Lock-In.
By the time that sunset clause hits, the most powerful AI firms will be deeply embedded in infrastructure, labor markets, government systems, and public norms. They won’t be regulated—they’ll be relied upon.
That’s the bet: That self-regulation by industry insiders will outperform institutions designed—however imperfectly—to represent the peoples’ will.
Yes, democracy can be slow, fragmented, even captured. But if public oversight fails, the answer isn’t surrender; it’s redesign.
Delegating self-governance purely to the private sector isn’t innovation. It’s abdication.
Citations:
Leave a Reply