White House orders federal preemption of state AI laws

The White House published an executive order on December 11, 2025 directing the Department of Justice to establish an AI Litigation Task Force whose sole responsibility is to challenge state AI laws that the administration considers inconsistent with national policy. The order, signed by President Donald Trump, represents the most aggressive federal preemption action on AI regulation to date.

The order explicitly names Colorado’s “algorithmic discrimination” law as a target, arguing it “may even force AI models to produce false results in order to avoid a ‘differential treatment or impact’ on protected groups.” It directs the Attorney General to form the Task Force within 30 days. The Commerce Secretary must publish an evaluation of existing state AI laws within 90 days, identifying those that “require AI models to alter their truthful outputs” or compel disclosures that would violate the First Amendment.

The administration is using federal funding as leverage. Section 5 of the order instructs the Commerce Department to issue a Policy Notice specifying that states with “onerous AI laws” are ineligible for non-deployment funds under the Broadband Equity Access and Deployment (BEAD) Program. The order also directs all executive agencies to assess their discretionary grant programs and consider conditioning grants on states not enacting conflicting AI laws.

This is not a neutral deregulatory gesture. The order picks a side in a structural fight that has been building since 2023, when states began passing their own AI laws in the absence of federal action. Colorado’s SB 24-205, signed in May 2024, requires developers and deployers of high-risk AI systems to conduct impact assessments and guard against algorithmic discrimination. Connecticut, California, and Texas have passed or proposed similar laws. The executive order treats every one of these as a potential threat to what it calls “United States’ global AI dominance.”

The order’s logic is straightforward: 50 state regimes create a compliance burden that hurts startups and large companies alike. The administration wants a “minimally burdensome national standard” that preempts state laws. But the order does not wait for Congress. It directs the FTC to issue a policy statement on whether state laws that require alterations to “truthful outputs” of AI models are preempted by the Federal Trade Commission Act’s prohibition on deceptive acts. It directs the FCC to initiate a proceeding on a federal reporting and disclosure standard for AI models that preempts conflicting state laws.

The legislative recommendation called for in Section 8 is notable for what it exempts from preemption. The order explicitly says the recommendation shall not propose preempting state laws relating to child safety protections, AI compute and data center infrastructure (excluding permitting reforms), and state government procurement and use of AI. This carves out space for states to regulate physical infrastructure and public-sector deployment while stripping them of authority over model behavior and disclosure.

The order frames state regulation as both a constitutional problem and a competitive one. It argues that state laws “impermissibly regulate beyond State borders, impinging on interstate commerce.” This is the dormant commerce clause argument that legal scholars have been watching for years. The DOJ’s AI Litigation Task Force will test this argument in court. The first target is almost certainly Colorado’s law, which takes effect in February 2026.

For AI companies, the order is a double-edged sword. A single national standard would reduce compliance costs. But the order’s aggressive preemption strategy creates legal uncertainty. Companies operating in Colorado, Connecticut, or California now face the prospect of being caught between state enforcement actions and federal litigation challenging those same laws. The BEAD funding threat adds a layer of pressure on state legislatures.

The order’s focus on “truthful outputs” deserves scrutiny. It treats any state law that might alter an AI model’s output as a form of compelled speech that violates the First Amendment. This is a contested legal theory. Courts have not yet ruled on whether requiring an AI model to avoid discriminatory outputs constitutes compelled speech in the constitutional sense. The FTC policy statement ordered in Section 7 will attempt to establish this position as federal policy.

The order does not address the substance of what constitutes a “truthful” output from an AI model. Models produce probabilistic predictions, not factual statements. A model that refuses to generate a biased output is not necessarily producing a false one. The Colorado law requires developers to take “reasonable care” to avoid algorithmic discrimination. The executive order treats this as a mandate to produce false results. The gap between these two interpretations is where the litigation will happen.

The European Union’s AI Act, which takes full effect in 2026, imposes obligations on high-risk AI systems including impact assessments and human oversight. The U.S. executive order moves in the opposite direction, preempting state-level risk management requirements. This creates a regulatory divergence that multinational AI companies will have to navigate.

The order’s timing matters. It was signed in December 2025, during the transition between administrations. The incoming administration inherits an AI Litigation Task Force, a Commerce Department evaluation, and an FTC policy statement in progress. The question is whether the next president will continue this preemption strategy or reverse it.

The order itself acknowledges that legislation is the preferred path. Section 8 calls for a “legislative recommendation establishing a uniform Federal policy framework for AI that preempts State AI laws.” But the order does not wait for Congress. It uses executive authority to challenge state laws, condition federal funding, and direct independent agencies to act. This is preemption by administrative action.

For AI builders, the message is clear: the federal government will fight state regulation in court and through the purse. The litigation task force is the mechanism. The BEAD program is the leverage. The FTC and FCC are the tools. Whether this strategy survives judicial review is an open question. What is certain is that the next phase of AI regulation will be fought in federal courtrooms, not state legislatures.