🎧 LISTEN TO THIS ARTICLE

The Trump administration found a creative use for broadband funding: killing state AI laws. An executive order signed December 11, 2025 instructs the Commerce Department to condition up to $42 billion in BEAD broadband grants on states repealing AI regulations the White House considers "onerous." States that passed AI bias auditing requirements or algorithmic transparency laws now face a choice: keep the rules or keep the money.

The order set a 90-day clock. By March 11, 2026, the FTC must issue a policy statement classifying state-mandated bias mitigation as a "per se deceptive trade practice." The Commerce Department must publish an evaluation identifying which state laws conflict with federal policy. An AI Litigation Task Force, stood up by January 10, was directed to challenge those laws in court.

All this while Congress has passed exactly zero comprehensive AI laws. The executive branch is building a national AI policy framework through executive order because the legislative branch won't.

1,208 Bills, Zero Federal Law

State legislatures haven't been waiting. In 2025 alone, lawmakers in all 50 states introduced 1,208 AI-related bills. Twenty-seven states enacted 73 new AI laws. California leads with 24 AI-related laws across 2024 and 2025. Colorado's AI Act requires developers to assess discriminatory impacts of high-risk AI systems. NYC's Local Law 144 mandates independent bias audits for automated employment decision tools. Illinois requires employers to notify candidates when AI analyzes their video interviews.

These aren't fringe regulations. They address hiring discrimination, algorithmic bias in lending, and automated decision-making that affects millions of people. They passed because Congress didn't act, and constituents demanded their states do something.

The executive order treats this activity as the problem. It frames state AI laws as obstructing "America's AI innovation, leadership, and global dominance." The word "safety" appears in the exceptions list. The word "dominance" appears in the mission statement. The priorities are clear.

The BEAD Gambit

The $42 billion Broadband Equity, Access and Deployment program was appropriated under the Infrastructure Investment and Jobs Act to expand internet access to underserved communities. The executive order repurposes this funding as a compliance tool. States identified as having "onerous" AI laws lose eligibility for non-deployment BEAD funds.

This mechanism is legally aggressive. Using federal funding to coerce state policy decisions triggers constitutional scrutiny under the Spending Clause. The Center for American Progress called the order "an unprecedented, unconstitutional, and dangerous assertion of the federal executive branch into the powers of state and local government." Georgetown's Center for Security and Emerging Technology concluded that "the moratorium may be more of a political liability than an innovation booster."

Colorado Attorney General Phil Weiser committed to challenging the order in court. California officials made similar statements. A bipartisan coalition of 40 state attorneys general had already written to Congress in May 2025 opposing federal preemption of state AI oversight. By November, a separate letter from 36 AGs opposed similar language in the NDAA. The coalitions included attorneys general from red and blue states alike.

The legal challenges are coming. The constitutional arguments against executive preemption of state law without Congressional authorization are strong. But litigation takes years. The regulatory chill starts immediately.

The FTC Flip

The March 11 FTC directive might be the most consequential element. The order instructs the FTC to issue a policy statement explaining when state laws requiring "alterations to the truthful outputs" of AI models constitute deceptive trade practices under Section 5 of the FTC Act.

Read that carefully. The executive order reframes algorithmic bias auditing as forcing AI to lie. If a state law requires a hiring algorithm to be audited for racial bias, and the audit leads to adjustments in the model's outputs, the White House now considers that a "deceptive practice."

This reverses years of FTC guidance. The agency's own prior statements positioned algorithmic bias as the liability risk, not the cure. Under the new framework, the cure becomes the liability. Companies that comply with state bias mitigation laws could theoretically face federal enforcement for doing so.

The practical result: companies building AI systems for regulated industries face irreconcilable mandates. Comply with Colorado's bias auditing requirement and risk federal enforcement. Ignore it and risk state enforcement. The legal uncertainty is itself the point. It paralyzes state-level regulation without formally repealing a single law.

Two Continents, Opposite Directions

While the US dismantles state AI guardrails, the EU AI Act reaches full enforcement on August 2, 2026 for most high-risk AI systems. Fines run up to 35 million euros or 7% of global revenue, whichever is higher. The EU framework requires risk assessments, bias documentation, human oversight for high-risk applications, and conformity assessments before deployment.

Any company building AI agents for international customers now operates under two fundamentally incompatible regulatory philosophies. The US says: don't let states mandate bias testing, it's deceptive. The EU says: you must mandate bias testing, it's required. A multinational deploying hiring agents in both New York and Berlin can't satisfy both regimes simultaneously.

We covered the EU AI Act's compliance requirements in our EU AI Act 2026 guide. The Trump executive order makes that guide more urgent, because companies can no longer assume US regulation will eventually converge with European standards. The two systems are heading in opposite directions.

What Happens Next

The March 11 deadlines are here. The Commerce Department's evaluation of "onerous" state AI laws will name specific targets. The FTC's policy statement will set the legal framework for challenging state bias mitigation laws at the federal level. Both will face immediate legal challenges.

The most likely near-term outcome: years of litigation while companies operate in regulatory limbo. States won't voluntarily repeal their AI laws to protect BEAD funding. The federal government lacks the legal tools to force repeal without Congressional action. Courts will eventually decide the constitutional questions, but "eventually" means 2028 or later.

For anyone building AI agents today, the practical takeaway is uncomfortable. Build for the strictest jurisdiction you operate in, because that's the only approach that survives all outcomes. If the Trump order holds, you lose nothing by having documented your bias testing. If it falls, you'll need those documents to comply with the state laws that survived.

The irony is hard to miss. An order designed to free AI from regulation has created more regulatory uncertainty than any single state law ever did. Companies that wanted clear rules now have three competing frameworks: state, federal, and European. The innovation the order claims to protect is exactly what regulatory chaos undermines.

Sources

Government:

Legal Analysis:

State Legislation:

Related Swarm Signal Coverage: