• What the states built
• The constitutional gambit
• Who benefits
• The $42 billion lever
• What happens in March

On December 11, 2025, President Trump signed an executive order titled “Ensuring a National Policy Framework for Artificial Intelligence.” The title sounds like bureaucratic tedium. The mechanism is a constitutional weapon.

The order creates a Department of Justice litigation task force, launched January 10, 2026, with a single purpose: to identify and sue states whose AI regulations the federal government considers overreaching. The constitutional basis is the dormant Commerce Clause — the same legal theory used to strike down state attempts to regulate railroads in the 19th century.

At least 38 state AI laws are being targeted. The Commerce Department must identify which ones deserve referral to the litigation task force by March 11. The FTC must issue a policy statement on AI preemption authority by the same date.

And here’s the detail that makes the mechanism something more than litigation: the order conditions $42 billion in broadband infrastructure funding on states repealing AI regulations deemed “onerous.” Comply, or lose your money.

§What the states built

While Congress spent years debating AI legislation without passing anything, states filled the vacuum.

California’s SB 53, effective January 1, 2026, requires transparency reports before launching frontier AI models — specifically those trained on computing power exceeding 10^26 FLOPs. Penalties run up to $1 million per violation. Texas passed the TRAIGA, creating an intent-based discrimination liability framework for AI, restricting government biometric identification, and establishing a regulatory sandbox. Illinois enacted HB 3773, focused on employment decisions made by AI systems.

By early 2026, the pipeline was accelerating. Oregon’s SB 1546 passed the Senate 26-1 on February 19, requiring AI chatbot disclosure and mental health crisis protocols. Washington had six AI bills survive a critical procedural deadline. Florida introduced an AI Bill of Rights that would prohibit government contracts with certain AI companies. California introduced the Digital Dignity Act to address AI-generated impersonation. Alabama created a study commission. Tennessee proposed making it a felony to train AI systems that encourage suicide.

This is not a handful of outlier states pursuing fringe regulation. This is a nationwide wave of legislation addressing AI in employment, healthcare, child safety, transparency, and civil rights. And the federal government’s response is not to pass its own laws. It’s to stop the states from passing theirs.

§The constitutional gambit

Senator Marsha Blackburn’s proposed TRUMP AMERICA AI Act represents the legislative arm of the strategy. It would codify the December executive order into law, creating a comprehensive federal framework that explicitly preempts state regulations.

The legal theory is straightforward: AI systems operate across state lines. Allowing 50 different regulatory frameworks creates compliance chaos. Federal uniformity serves interstate commerce. This was the same argument that consolidated railroad regulation in the 1880s, telecommunications regulation in the 1930s, and internet regulation in the 1990s.

The problem is that the President cannot unilaterally declare preemption. Executive orders can direct agencies to challenge laws in court, but binding preemption requires either an act of Congress or court rulings invalidating specific state provisions. The executive order is a declaration of intent, not a legal conclusion.

Legal scholars note that the cases could reach the Supreme Court. The constitutional question — whether AI regulation falls within the federal government’s exclusive commerce authority — has no precedent. States will argue that AI impacts health, safety, and welfare, which are traditionally state domains. The federal government will argue that AI is inherently interstate and that fragmented regulation stifles innovation.

Twenty-four state attorneys general sent a letter to the FCC on December 19, opposing federal preemption. That’s nearly half the states, signaling organized resistance.

§Who benefits

Major technology industry associations have publicly supported federal preemption. Their argument is consistent: uniform standards reduce compliance costs and prevent regulatory fragmentation. A company deploying an AI system nationally shouldn’t need 50 different compliance frameworks.

This argument is true. It also happens to mean that whatever the federal standard becomes — including no standard at all — replaces everything that states have built.

The executive order exempts four categories from preemption: child safety protections, data center infrastructure rules, state government procurement requirements, and AI transparency for public sector systems. Everything else is fair game. Employment discrimination protections. Healthcare AI standards. Biometric surveillance limits. Civil rights safeguards. If a state enacted it and the federal government considers it burdensome, it’s a target.

Hundreds of civil society, labor, and consumer protection organizations have opposed preemption, warning that it could undermine transparency, accountability, and civil rights protections. Their concern is specific: preemption doesn’t just replace state laws with federal ones. It can replace state laws with nothing.

§The $42 billion lever

The broadband funding condition is the most operationally significant element. States that have accepted federal broadband infrastructure money now face a choice: repeal AI regulations, or risk losing funding for an entirely different policy priority.

This kind of cross-conditional funding is not new. The federal government has used highway funding to enforce drinking age laws and education funding to enforce civil rights requirements. But applying it to AI regulation represents something different — using telecommunications infrastructure money to control technology governance.

For state legislators who passed AI regulations over the last two years, the calculation has changed. The question is no longer whether their laws are good policy. It’s whether they can afford them.

§What happens in March

The March 11 deadlines are the next inflection points. The Commerce Department’s identification of “burdensome” state laws will be the first official signal of which regulations the federal government intends to challenge. The FTC’s preemption policy statement will reveal how aggressive the federal interpretation will be.

Between now and then, states are still passing laws. Oregon’s chatbot safety bill is advancing. Washington’s bills are crossing chambers. California has multiple AI proposals in committee. These legislators are building on sand that the federal government may be about to wash away.

The preemption playbook treats state AI regulation as a problem to be solved, not a democratic process to be respected. Whether that gambit succeeds will depend on courts, Congress, and the willingness of state governments to fight for regulatory authority they assumed they had.

The constitutional question underneath all of this is simple, even if the answer isn’t: who gets to decide what AI can do to people?

Executive order details from the Federal Register, December 11, 2025. State legislation from the Transparency Coalition AI Legislative Update, February 20, 2026. Constitutional analysis from the National Law Review (Jones Walker LLP), King & Spalding, Mintz, and Baker Botts analyses. State attorney general letter from December 19, 2025, as reported by multiple outlets.

Originally published at https://noahaust2.github.io/strategist-dashboard/blog/the-preemption-playbook.html