Representatives Jay Obernolte, Republican of California, and Lori Trahan, Democrat of Massachusetts, on Thursday released a 269-page discussion draft of the Great American Artificial Intelligence Act of 2026, a bipartisan proposal that would establish the first comprehensive federal A.I. governance framework and, for three years, preempt state laws regulating the development of A.I. models. The draft, co-sponsored by Representatives Scott Franklin, Suhas Subramanyam, Erin Houchin, and Scott Peters, lands as the most ambitious congressional attempt yet to convert what has been a state-led regulatory experiment into a single federal regime.

The bill would codify the Center for AI Standards and Innovation, the Commerce Department body that Secretary Howard Lutnick rebranded last June from the Biden-era U.S. AI Safety Institute. CASI would receive $100 million per fiscal year, coordinating with the Energy Department and the National Institute of Standards and Technology. Frontier developers with more than $500 million in prior-year gross revenue would be required to publish safety frameworks and report critical safety incidents, according to Roll Call, with penalties authorized at $1 million per day per violation, well above the $1 million-to-$3 million per-incident caps now common in state law.

The preemption clause is what’s generating the fight. An accompanying document released by Ms. Trahan’s office concedes that California’s AB 2013 training-data disclosure law and the watermarking portion of SB 942 would be displaced, while frontier safety statutes in California, New York and Illinois would be “federalized.” The International Association of Privacy Professionals notes the bill builds on Illinois’s recently passed SB 315 by expanding third-party audits.

Obernolte framed the trade-off in geopolitical terms: “America should lead the world in artificial intelligence, not regulate ourselves into falling behind China through a patchwork of fifty different state laws.”

Safety advocates aren’t buying it. Brad Carson, president of Americans for Responsible Innovation and a former Democratic representative from Oklahoma, called the preemption provision a “generational mistake.” “This bill takes the current floor on state AI legislation and turns it into a federal ceiling, preventing state lawmakers from addressing emerging AI harms in an era of fast-moving technology,” he said. Brendan Steinhauser, chief executive of the Alliance for Secure AI, was blunter: “A national AI standard should protect at least as much as it preempts.” Industry took the opposite read. Jason Oxman, chief executive of the Information Technology Industry Council, praised “key elements that will accelerate American AI leadership.”

The draft arrived two days after President Trump signed an executive order establishing voluntary federal agency reviews of new frontier models, according to FedScoop, putting two parallel governance theories on the table in the same week: voluntary executive review and statutory federal preemption. The 1996 Telecommunications Act produced a similar consolidation moment for the internet’s first regulatory decade, and the durability of that bargain depended less on what the law said than on which constituencies were allowed to keep legislating around it. Three years is a short ceiling. It’s also long enough to set the defaults.

Sources