Federal preemption is the legal principle that when a valid federal law conflicts with a state law, the federal law wins and the state rule has to give way. It comes from the U.S. Constitution’s Supremacy Clause, and it has become the central legal strategy behind a push to override a growing patchwork of U.S. state artificial-intelligence laws — from Colorado’s rules on “algorithmic discrimination” to dozens of other state AI bills passed over the past two years.

What federal preemption actually means

Federal preemption is not new and has nothing to do with AI specifically — courts have used it for decades in areas like food labeling, airline pricing, and auto-safety standards. Lawyers generally group it into three types: express preemption, where a federal statute states outright that it overrides state law; field preemption, where federal regulation of an area is so complete that no room is left for states; and conflict preemption, where complying with both federal and state rules at once is impossible, or a state law blocks what a federal law is trying to achieve. Courts normally start from the assumption that Congress did not mean to wipe out state law unless it said so clearly — preemption is the exception, not the default.

One wrinkle matters for AI: preemption is a product of federal statutes, not presidential wishes. A president cannot simply declare state laws void by executive order. What an order can do is direct federal agencies to use the powers Congress already gave them — through lawsuits, regulations, and funding rules — to make the preemption case.

Why AI became a preemption battleground

In the absence of a single federal AI law, more than a dozen U.S. states passed their own rules covering topics like algorithmic discrimination in hiring and lending, chatbot disclosure, and deepfakes. Colorado’s law restricting algorithmic discrimination — covered in our explainer on the Colorado AI Act — became a frequent target for critics who argue that a 50-state patchwork of different compliance rules is unworkable for companies operating nationwide.

That argument fed directly into Executive Order 14365, “Ensuring a National Policy Framework for Artificial Intelligence,” which President Trump signed on December 11, 2025. Rather than trying to nullify state laws outright, the order instructs federal agencies to pursue preemption through the legal tools already available to them.

The three levers being used

  • Litigation. The order directed the Department of Justice to stand up an AI Litigation Task Force, created in January 2026, whose sole job is suing states over AI laws the administration considers unconstitutional interference with interstate commerce or already preempted by federal regulation.
  • Agency rulemaking. The Federal Trade Commission was told to explain how Section 5 of the Federal Trade Commission Act — which bars deceptive business practices — could preempt state laws that require companies to alter an AI model’s outputs. The FTC’s proposal, covered in our news report, singles out Colorado’s law and offers companies a safe harbor: disclose clearly that a system is tuned toward something other than raw accuracy, and the disclosure shields the company from a deception claim.
  • Funding leverage. States whose AI laws the Commerce Department judges “onerous” can be cut off from remaining federal broadband funding under the BEAD program, and other agencies were told to consider similar conditions on discretionary grants.

The order also carves out limits: state laws on child-safety protections, AI compute and data-center infrastructure, and a state government’s own procurement and use of AI are explicitly excluded from this preemption push.

What it means for businesses and consumers

Because an executive order cannot by itself cancel a validly enacted state law, nothing here is settled yet. What Executive Order 14365 produced is a strategy — a litigation campaign, agency proposals still open for public comment, and funding threats — not a final legal outcome. Actual preemption will ultimately be decided by federal courts, or by Congress if it passes nationwide AI legislation, as the order itself calls for. Until then, companies operating across state lines still need to track each state’s requirements even as the legal ground shifts under them.

In the news

The FTC’s proposed policy statement, which is open for public comment, is a direct product of this strategy — see our report on the FTC’s proposal to preempt state AI bias laws.

FAQ

Does the executive order automatically cancel state AI laws? No. It directs federal agencies to challenge state laws in court, through regulation, or by conditioning funding — it cannot unilaterally invalidate a state law on its own.

Which state AI laws are being targeted? Colorado’s algorithmic-discrimination law is the most frequently cited example, but the strategy is aimed broadly at state AI rules the administration views as conflicting with federal policy.

Are any state AI laws protected from this push? Yes — the order explicitly excludes state laws on child safety, AI compute and data-center infrastructure, and government procurement of AI.

What should a business do in the meantime? Continue complying with applicable state law while monitoring the FTC’s rulemaking and the Justice Department’s litigation, since none of these preemption efforts have been finalized by a court or Congress.