FTC Floats a Policy Statement on AI 'Accuracy' That Could Preempt State AI Laws

The Federal Trade Commission has waded into the escalating fight between Washington and the states over who gets to regulate artificial intelligence, and it has done so with an argument that could reshape the terrain: that a state law telling AI companies how to shape their models' outputs may be quietly unlawful under federal law.

In a proposed policy statement released July 1 and opened for public comment through July 31, the FTC contends that companies which "distort their systems' outputs to achieve undisclosed ideological objectives" may be deceiving consumers in violation of Section 5 of the FTC Act, the agency's core prohibition on "unfair or deceptive" business practices. Woven into that consumer-protection theory is a more consequential legal claim aimed squarely at the states. The Commission singles out Colorado's Artificial Intelligence Act as a law that "appears to coerce companies into altering the output of their AI models to comply with and advance the state's ideological objectives," and asserts that such a law is "impliedly preempted to the extent it conflicts with a federal regulatory scheme."

That single sentence is why the document is drawing outsized attention. A policy statement is not a rule and carries no independent force of law, but it signals how the agency intends to read its own authority, and here the FTC is positioning Section 5 as a federal scheme capable of displacing state AI legislation.

From executive order to agency action

The statement did not emerge in a vacuum. It is a direct deliverable of President Trump's December 11, 2025 executive order, "Ensuring a National Policy Framework for Artificial Intelligence," which set out to build what the administration calls a "minimally burdensome" national approach and to dismantle what it views as a conflicting patchwork of state rules. The order directed the FTC to issue a policy statement addressing the legal implications of state laws that require alteration of the "truthful outputs of AI models." It also created a Department of Justice AI Litigation Task Force to challenge state AI laws in court, and directed the Commerce Department to weigh withholding federal broadband funds from states whose AI laws conflict with national priorities.

The FTC framed its move as advancing that agenda explicitly. "The FTC wants to hear from businesses and consumers about their experiences and concerns regarding the subversion of AI systems for ideological ends," Chairman Andrew N. Ferguson said in announcing the request for comment. "This crucial input will help the Commission formulate a final policy that advances President Donald Trump's goal of expanding America's global dominance in artificial intelligence."

The statement's consumer-protection logic rests on a chain of premises: that AI companies make explicit and implicit representations that their systems produce the most accurate output possible; that consumers rely on those representations, accepting AI answers without independent fact-checking the vast majority of the time; and that secretly engineering a model to serve undisclosed goals therefore misleads the public. From there, the FTC argues, a state law that pressures companies into that very engineering, in Colorado's case to avoid "disparate impact" liability, collides with the federal regime.

A contested legal theory

Preemption arguments of this kind are far from a sure thing. Implied preemption typically requires either that federal law occupies a field entirely or that state and federal requirements genuinely conflict. Critics note that Section 5 is an enforcement authority, not a comprehensive regulatory framework of the sort courts have historically found preemptive, and that Congress has repeatedly declined to pass federal AI legislation that would preempt the states outright. Colorado, for its part, has already moved: Governor Jared Polis signed Senate Bill 26-189 in May, repealing the original AI Act and replacing it with a narrower disclosure-and-rights framework set to take effect in January 2027, after a federal court paused enforcement of the earlier law in April.

The politics are unavoidable. The framing around "ideological" manipulation and "subversion" of AI mirrors a broader administration campaign against what it characterizes as biased or "woke" AI, and civil-liberties and consumer groups are likely to argue that the government invoking deception law to police model behavior raises its own First Amendment and federalism questions. Supporters counter that a fragmented, state-by-state regime is unworkable for a national technology and that consumers deserve to know when outputs are being shaped by undisclosed agendas.

It is worth noting the statement's modest institutional footing: the vote authorizing the Federal Register notice was 2-0, reflecting an unusually thin Commission, and comments will run through the end of July before any final version is adopted.

What to watch next

The near-term signal will be the comment record itself, particularly whether major AI developers, state attorneys general, and advocacy groups line up for or against the preemption theory. Longer term, the real test comes if the FTC or the DOJ task force tries to convert this reasoning into an enforcement action or a court filing, where a judge, not the agency, decides whether Section 5 can displace a state statute. Watch also for parallel moves at the FCC on federal disclosure standards and for the Commerce Department's funding evaluations. Together they will reveal whether Washington's push for a single national AI framework advances through persuasion, litigation, or leverage, and whether the states push back.