States Forge Ahead on AI Rules, Defying Trump's Push to Federalize Regulation

Six months after President Trump warned states to get out of the business of regulating artificial intelligence, a striking number of them are doing exactly the opposite. From statehouses in Sacramento and Denver to Albany and Tallahassee, lawmakers of both parties are pressing ahead with their own AI rules — setting up a federal-versus-state collision that is now playing out in courtrooms, executive orders, and an increasingly tangled patchwork of laws taking effect this summer.

The standoff crystallized this week. Reporting on June 14, 2026 confirmed what compliance lawyers had been bracing for: despite a December 2025 executive order directing the Justice Department to challenge state AI statutes, and despite a dedicated DOJ "AI Litigation Task Force" stood up in January, states are not backing down. A mix of Republican- and Democratic-led states — including Colorado, Connecticut, Idaho, Iowa, Nebraska, and Oregon — have passed AI legislation this year, much of it focused on disclosing when a person is interacting with a machine, protecting children from chatbots, and policing AI use by employers.

The federal push to centralize

The Trump administration's position is that a fifty-state mosaic of conflicting rules will smother American AI development and hand an advantage to China. On December 11, 2025, the president signed an executive order, "Ensuring a National Policy Framework for Artificial Intelligence," calling for a "minimally burdensome" national standard. The order instructed the attorney general to create an AI Litigation Task Force within 30 days to challenge state laws on theories including unconstitutional burdens on interstate commerce and preemption by existing federal statutes. It also floated using leverage — conditioning federal broadband funding on a state's policy alignment — to discourage state action.

The order named names. It singled out California's data-transparency requirements for generative AI and Colorado's algorithmic-discrimination law as the kind of "onerous" rules Washington intends to fight. The legal industry took the threat seriously; firms from Mayer Brown to Latham & Watkins published client alerts warning of a coming wave of preemption litigation.

The states dig in

The states, so far, are unmoved — and the resistance is bipartisan. In Florida, Governor Ron DeSantis dismissed the idea that Washington should hold a monopoly on AI rules while producing none of its own. "Preempting states re: AI without enacting a sensible federal framework is just an amnesty for Big Tech," DeSantis wrote on X, casting the administration's posture as a giveaway rather than a guardrail.

California has become the central battleground. Its AB 2013, the Generative Artificial Intelligence Training Data Transparency Act, took effect January 1, 2026, requiring developers to disclose the data behind their models. Elon Musk's xAI sued the state on December 29, 2025, arguing the law compels speech in violation of the First Amendment and exposes trade secrets behind its Grok chatbot. On March 4, 2026, a federal judge denied xAI's bid for a preliminary injunction, finding the company unlikely to succeed on the merits. Notably, rivals OpenAI and Anthropic did not sue and have already posted their AB 2013 disclosures. California lawmakers, undeterred, are now advancing the "No Robo Bosses Act of 2026," which would bar employers from relying solely on AI to fire or discipline workers.

The momentum extends well beyond California. New York's Governor Kathy Hochul finalized the Responsible AI Safety and Education (RAISE) Act, which targets frontier-model developers with at least $500 million in revenue and takes effect January 1, 2027. And Colorado — long the bellwether for comprehensive AI regulation — illustrates how messy the picture has become. Its landmark 2024 AI Act was set to take effect June 30, 2026, but on May 14 Governor Jared Polis signed a repeal-and-replace measure, the Colorado Automated Decision-Making Technology in Consequential Decisions Act, swapping the broad framework for a narrower one. Even where states are scaling back, they are still regulating.

The preemption problem

At the heart of the fight is a hard constitutional question: can the executive branch preempt state law on its own? Legal scholars are skeptical. Preemption ordinarily flows from Congress, not from an executive order, and the early courtroom signals — the denial of xAI's injunction chief among them — suggest judges are not inclined to wave state AI laws aside on commerce-clause or First Amendment grounds without a stronger record. The administration's strongest play may be Congress, but federal AI legislation has repeatedly stalled; a proposed preemption rider was stripped from the FY 2026 defense authorization bill after bipartisan pushback that included DeSantis.

That leaves industry caught in the middle. AI developers genuinely fear the compliance burden of a patchwork — different disclosure rules in California, different liability triggers in New York, different definitions of "high-risk" in Colorado. But their responses have splintered. Some, like xAI, are litigating. Others, like OpenAI and Anthropic, are quietly complying while lobbying for a single federal standard they can live with. The divergence underscores that "the industry" does not speak with one voice on whether Washington or the states should be in charge.

What to watch next

Colorado's narrowed AI law and the broader compliance deadlines arriving June 30 will be the next stress test of how far states will go in practice. Watch the DOJ task force for its first formal lawsuits — none had landed as of mid-June, and an empty docket would itself be telling. Watch the appeal in xAI v. Bonta, which could become the leading precedent on whether AI-transparency mandates survive constitutional scrutiny. And watch Congress: the administration's surest path to preemption runs through a federal statute it has not yet been able to pass. Until it does, the states appear content to fill the vacuum — and to dare Washington to stop them.