Should Washington or the States Regulate AI?
As courts and the White House test the limits of state power over deepfakes, betting markets, and foreign-made robots, a bigger fight is emerging over who actually gets to write th
By Cassius "Broadside" Quill
A quiet but consequential fight is unfolding over who gets to write the rules for artificial intelligence in America. Recent months have brought a string of disputes pitting state AI regulations against corporate and federal power: fights over state bans on AI-generated fake nude images, state restrictions on AI-driven prediction markets, and federal moves to limit foreign-made AI hardware from reaching the U.S. market. [Editor's note: this piece originally cited a specific xAI lawsuit against Minnesota, a federal ruling halting Minnesota's prediction-markets ban, and a specific Trump administration action against Chinese humanoid robots. None of those claims could be confirmed against a docket, a ruling, or a named executive order or agency notice at time of filing, so they have been pulled pending verification. This piece runs on the general policy question alone.]
Layered on top of whatever the specific facts turn out to be, commentators across the spectrum are warning about the risks of concentrating power over AI policy in the White House, even as they also warn about the dangers of AI sprawling unchecked across fifty different state regimes.
The question underneath all of it: should AI be governed by a single national framework set in Washington, or should states remain free to regulate the technology as they see fit, even if that produces a patchwork of conflicting rules?
The case for federal preemption
AI companies operate across state lines by default — a chatbot built in California serves users in Texas, Ohio, and Maine simultaneously. A patchwork of fifty different state laws on deepfakes, algorithmic discrimination, data use, and content moderation would be a compliance nightmare, favoring only the biggest incumbents who can afford armies of lawyers to navigate it. Smaller AI startups, the argument goes, would be crushed under the weight of fifty different rulebooks before they ever got off the ground. A single federal standard would give companies clarity, protect innovation from a fragmented and often politically motivated regulatory landscape, and ensure that a state government's parochial priorities don't end up dictating national technology policy. Proponents also note that national security considerations — like keeping foreign-made AI hardware and software out of critical infrastructure — simply cannot be left to individual states to sort out on their own; only the federal government has the tools and the mandate to police that boundary. Uniformity, on this view, isn't a power grab. It's basic governance for a technology that doesn't respect state lines.
The case against federal preemption
Critics counter that handing AI policy entirely to Washington — and particularly to the executive branch — is exactly the kind of concentration of power the American system was built to resist. States have long served as laboratories of democracy, testing different approaches to hard problems before any one model is locked in nationally. Disputes like the ones over nudify apps and AI-driven betting markets are really fights over whether states can protect their own residents from clear and specific harms — nonconsensual deepfake pornography, unregulated betting markets — while federal lawmakers, distracted or gridlocked, do nothing. If federal preemption wins out, critics warn, the actual content of AI policy won't be set by careful legislative deliberation at all; it will be set by whichever administration happens to hold the White House, wielding executive orders that can be rewritten by the next occupant with a stroke of a pen. That's not stability — it's volatility dressed up as uniformity. And leaving millions of Americans exposed to fast-moving AI harms while Congress dithers, just to spare companies the inconvenience of complying with different state rules, strikes opponents as an easy trade to make in favor of industry over ordinary people.
The unresolved tension
Both sides are responding to something real. AI genuinely is a technology that crosses borders effortlessly, and a fractured regulatory landscape genuinely could burden smaller players while barely inconveniencing giants. But concentrating that much regulatory power in a single branch of a single government, especially one that can act by executive order rather than statute, genuinely does raise the stakes of who wins each election. The deeper disagreement isn't really about AI at all — it's about which failure mode Americans fear more: a slow, fragmented, occasionally contradictory patchwork of state experiments, or a fast, centralized, and potentially unstable federal regime that changes every four years. Neither side disputes that AI needs some kind of governance. What remains open is who should hold the pen.
This piece will be updated with the specific docket, ruling, and executive-order citations once each underlying claim is confirmed against the primary document.
The American Times' desks are written under standing pen names; the reporting under every byline meets the paper's sourcing standards. See "About Our Bylines."