This coverage has examined how the European Union builds a centralized framework to regulate artificial intelligence and how most of the world watches the major powers set the rules. The United States offers a revealing counter-case: not a debate over how to regulate AI, but over who gets to — the federal government or the fifty states. It is a structural fight about power and federalism, and in 2026 it has produced a standoff that says as much about the American system as about the technology.
Begin with the action that crystallized the conflict. On December 11, 2025, the president signed Executive Order 14365, titled “Ensuring a National Policy Framework for Artificial Intelligence.” Its stated purpose is to sustain US “global AI dominance” through what it calls a “minimally burdensome national policy framework” — and, in the meantime, to “check the most onerous and excessive laws emerging from the States.” The order directs federal agencies — Justice, Commerce, the FTC, the FCC — to identify state AI laws the administration deems inconsistent with federal priorities, threatens states that enact “onerous” rules with the loss of certain federal funding, and creates an AI Litigation Task Force within the Department of Justice to challenge those laws in court starting in January 2026.
The catch: an order is not a law
Here is the legal-technical core, and it is the hinge of the entire story. Federal preemption — the power of federal law to override state law — typically flows from an act of Congress, not from an executive order. As multiple legal analyses stress, the order would likely not, on its own, displace state AI laws; it functions as guidance for federal agencies and a statement of policy, not as a self-executing override. Preemption, in the analysts’ phrase, is not automatic. To make it stick, the administration needs either favorable court rulings or, more durably, legislation from Congress.
And that is precisely what the administration has not been able to get. The order is the latest move in a sustained effort that Congress has twice declined to ratify. A proposed ten-year moratorium on state AI enforcement — later narrowed to five — was stripped from a major budget bill in mid-2025 when the Senate voted nearly unanimously, 99 to 1, to remove it. Lawmakers then declined again to include state-preemption language in the 2026 National Defense Authorization Act, despite a direct push. The executive order, in effect, takes up the baton from these failed legislative attempts — which is also why its legal durability is uncertain.
A genuinely bipartisan split
What makes this fight unusual, and worth understanding, is that it does not divide cleanly along party lines. The opposition to federal preemption is strikingly bipartisan. Some Republican governors and senators have objected loudly: one prominent governor called the moratorium “a subsidy to Big Tech” that “lets technology companies run wild,” and others joined Democrats in resisting it. Congressional Republicans who support at least some state-level AI regulation have been among the obstacles. On the other side, a Senate committee chair has been a vocal champion of preemption, aligned with the White House.
The defense of state authority, too, crosses the aisle: a Democratic governor warned the order “overrides important state protections,” while a Republican governor framed the same measure as a giveaway to large technology firms. This is not the usual left-right axis; it is a contest between those who prioritize a single, light-touch national standard to favor innovation and those who — for varied reasons — want states to keep their power to legislate. Notably, the final order was narrowed to expressly preserve certain state laws, including those on child safety, an acknowledgment of where the political resistance was fiercest.
What is actually at stake for the states
The concrete stakes are real laws already on the books or about to take effect. States have moved into the vacuum left by congressional inaction: Colorado enacted a comprehensive AI Act; California passed transparency requirements for frontier AI models, plus rules on companion chatbots and election deepfakes examined elsewhere in this coverage. These are the kinds of measures the federal effort targets as “onerous.” Yet even legal analysts skeptical of the states’ position doubt the order will eliminate state involvement: states are expected to keep regulating AI indirectly, applying existing consumer-protection, unfair-competition, deceptive-practices, and civil-rights laws to AI conduct. The genie of state regulation, once out, is hard to put back by executive fiat.
Two readings, with comparable weight
The conflict admits two legitimate positions, worth presenting without tilting the scale.
Those who favor federal preemption argue that a patchwork of fifty different state regimes is unworkable for a national technology: companies cannot build to fifty conflicting standards, the compliance burden falls hardest on smaller firms, and a single, predictable national rule is better for innovation and for US competitiveness against China. From this angle, the executive order is a reasonable attempt to prevent regulatory fragmentation from hobbling a strategic industry.
Those who defend state authority argue that, in the absence of any federal AI law, the states are the only governments actually protecting citizens — on algorithmic discrimination, deepfakes, child safety, automated decisions — and that a moratorium or preemption would leave a protection vacuum benefiting large technology companies. They add a constitutional point: that overriding state law by executive order, without congressional action, oversteps presidential authority and will likely be challenged in court. From this angle, the fight is about preserving the states’ traditional power to protect their residents when Washington will not.
It is not for this outlet to decree which side is right; the matter will be settled by courts and, ultimately, by whether Congress acts. What can be stated is that both describe a real tension: the genuine cost of fragmentation for a national industry, and the genuine risk of a protection vacuum if state rules vanish before any federal ones replace them.
What this case reveals
What the US preemption fight adds to the coverage is a different face of the regulation problem. Elsewhere the question was whether to regulate AI and how; here, uniquely, the technology is almost secondary to a prior question — who holds the authority to make the rules at all. It is a reminder that governing a borderless technology runs into not only the gap between law and innovation, but the internal architecture of each state: federal systems must first decide, among themselves, who governs, before they can govern. The contrast with the EU’s centralized approach is instructive — and the world is watching which model proves more workable, because both offer templates others may follow.
The verifiable fact is that the US executive sought to impose a national AI framework by overriding state laws, that Congress has twice refused to enact such preemption, that an executive order alone likely cannot displace state law, and that the result is an unresolved standoff in the world’s largest technology market. Whether the US ends up with a single national standard or a continuing mosaic of state rules will depend on decisions not yet made: on how courts rule on the litigation now beginning, on whether Congress finally legislates, and on whether states defend their authority or yield to federal pressure. As in every story of this kind, what is decisive is not the technology itself, but the prior question of who gets to govern it — a question the United States, for now, has not answered.