Even with all three branches of the federal government involved, preempting state AI laws will require bipartisan compromise.

Last week, Representatives Jay Obernolte (R-CA) and Lori Trahan (D-MA) released a draft of the Great American AI Act, which would create a nationwide regulatory framework for the technology. The bill is the latest attempt at federal preemption.

In November 2025, President Trump declared that the United States “MUST have one Federal Standard” for AI, “instead of a patchwork of 50 State Regulatory Regimes.” A month later, Trump issued an Executive Order on Ensuring a National Policy Framework for Artificial Intelligence, aiming to create a “minimally burdensome national standard—not 50 discordant State ones.”

The EO directed executive agencies to undertake a variety of preemption-focused initiatives by March 11. Only the Department of Justice met the deadline: on January 9, the DOJ established “an AI Litigation Task Force” to bring legal challenges against state AI laws that “unconstitutionally regulate interstate commerce, are preempted by existing Federal regulations, or are otherwise unlawful.”

The other agencies missed the mark. The Department of Commerce has yet to publish its evaluation of “onerous” state AI laws or a “Policy Notice” stipulating that states with such laws are ineligible for funding from the Broadband Equity Access and Deployment Program. The delay by Commerce has left the Federal Communications Commission in limbo: the EO instructed the FCC to “initiate a proceeding to determine whether to adopt a Federal reporting and disclosure standard for AI models that preempts conflicting State laws” within ninety days of Commerce releasing its evaluation.

The Federal Trade Commission also failed to “issue a policy statement” explaining the application of federal consumer protection law to AI and the extent to which the FTC’s deception authority preempts certain state laws. In April, the FTC claimed the “policy statement will come out very soon,” but the Commission is over two months late, with no statement in sight.

In April, the DOJ’s AI Litigation Task Force won its first legal challenge. On April 9, X.AI sued Colorado in federal court to block enforcement of the Colorado AI Act. Two weeks later, the DOJ intervened in opposition to the state law, and, that same day, the Colorado Attorney General agreed to stay, or pause, enforcement. Three days later, the court fully blocked enforcement.

Since then, Colorado lawmakers have advanced a bill to repeal and replace the Act with a narrower framework. Governor Polis is expected to sign the new bill into law, effectively killing the original statute.

Litigation alone, however, is not enough to prevent States from regulating AI. Lawsuits can only tackle one law at a time, and plaintiffs must prove challenged statutes are unconstitutional, likely resulting in complex, lengthy litigation. In Colorado, the DOJ relied on a private plaintiff, X.AI, to sue; the DOJ faces greater challenges in cases it brings on its own.

Now, the Trump administration appears to have changed its preemption strategy, shifting the focus to Congress—where it should be. In March, the White House released its National Policy Framework for Artificial Intelligence, calling on the legislative branch to enact federal statutes to “preempt state AI laws that impose undue burdens.”

Congressional preemption efforts started in June of last year, months before Trump’s EO, when Senator Cruz (R-TX) tried to insert a moratorium on state AI laws into the budget reconciliation bill. The Senate voted down the moratorium 99–1, killing legislative momentum.

The AI legislative framework restarted the process. In March, Senator Blackburn (R-TN) released a draft of the TRUMP AMERICA AI Act. The “RU” in TRUMP stands for “Republic Unifying”: the bill would preempt state laws that conflict with its provisions. Blackburn’s explicit goal is “to solve the patchwork of state laws.”

Senators Todd Young (R-IN) and Maria Cantwell (D-WA) both support a uniform federal standard but are taking a more subtle approach. Their bill, the Future of AI Innovation Act, would codify the Center for AI Standards and Innovation as the epicenter of federal authority, directing CAISI to assist federal agencies “in developing voluntary best practices” for AI systems “to best fulfill their [regulatory] missions.” By doing so, the Act would enable federal agencies to occupy the entire field of AI regulation, impliedly preempting state AI laws.

In their bill, Representatives Jay Obernatle (R-CA) and Lori Trahan (D-MA) are borrowing from both approaches: like Blackburn, they overtly call for “one national standard”; however, like Young and Cantwell, they are applying a softer touch, compromising on the scope and timeline of federal preemption. The Great American AI Act would “preempt State regulation specifically targeting artificial intelligence model development,” but only for three years after enactment. And the bill leaves intact state laws “of general applicability” as well as those governing “post-deployment activities,” including “implementation, deployment, distribution, offering, or use of any artificial intelligence system, product, or service.”

Despite these concessions and bipartisan backing, the bill is already facing pushback. The House Commission on AI and the Innovation Economy stated that it “cannot serve as the basis for productive dialogue,” with Co-Chair Valerie Foushee (D-NC) recently denouncing “any and all efforts to” preempt states. Even the private sector has started to sour on preemption, with Anthropic arguing it should be “construed narrowly, with ambiguity resolved in favor of preserving state authority.”

Federal oversight of AI is firmly anchored in both policy and constitutional law. But Congress, the only branch with the authority for broad preemption, is gridlocked. Despite the federal government’s coordinated approach and legislators’ growing willingness to compromise, States remain in the driver’s seat, and a regulatory patchwork seems inevitable.