AI Preemption Battle Lands in Congress With Substantive Discussion Draft

WASHINGTON, June 4, 2026 - A leading House Republican and Democrat have floated a 269-page discussion draft for a sweeping artificial intelligence bill that would freeze state laws on the topic for three years, and also force the country's most powerful frontier labs to open up their models.

Mooted on Thursday by Rep. Jay Obernolte , R-Calif., and Rep. Lori Trahan , D-Mass., the draft, formally titled the Great American Artificial Intelligence Act of 2026, would allow states to retain the power to regulate the use of AI systems within their borders. But states would lose the ability to legislate on how those systems are built.

See Breakfast Club Membership Options!

                        See Breakfast Club Membership Options
                    

That’s the part of the process that critics say matters most for safety.

The bill would place binding requirements on “large frontier developers,” the big AI companies with more than $500 million in annual revenue that have trained the most powerful AI models.

Such companies would be required to publish a detailed safety plan, known as a “frontier AI framework,” covering how they identify and manage the most severe risks their models could pose.

Precise definition of catastrophic risk

The bill defines catastrophic risk precisely: A foreseeable risk of death or serious injury to more than 50 people, or more than $1 billion in property damage, from a model helping develop a weapon of mass destruction, conducting cyber attack or taking harmful autonomous action without meaningful human oversight.

Large frontier labs would be required to hire licensed independent auditors every six months to check compliance with those safety plans. They would have full access to company records, personnel, and systems.

Labs that fail to comply could face civil penalties of up to $1 million per violation per day. Before releasing a new model, the labs must also publicly disclose what risks were assessed, the results, and what steps were taken in response.

Oversight in a Commerce Department agency

Oversight would sit with the Center for AI Standards and Innovation, a Commerce Department office that the bill would formally establish in law and fund with $300 million in over three years. The center already works voluntarily with OpenAI, Anthropic, and Google’s DeepMind.

The transparency and auditing requirements, along with the freeze on the broadest state powers, would expire automatically after three years unless renewed by Congress.

Beyond frontier AI governance, the bill includes whistleblower protections for AI employees, tougher criminal penalties for AI-assisted fraud, workforce research programs tracking AI’s impact on jobs, and education initiatives from K-12 literacy to university scholarships.

Obernolte, chair of Republican Policy Committee, gives the draft weight

Obernolte, chair of the House Republican Policy Committee and one of Congress's most prominent Republican voices on AI, gives the bill institutional weight. With a Democrat co-signing, it is Republicans' most viable vehicle for federal AI rules before the midterm elections.

The bill arrives as the administration's own preemption push has stalled in the courts. A December 2025 executive order directed the Commerce Department to withhold some of its $42.5 billion in Broadband Equity, Access and Deployment program funding from states with AI laws it deemed onerous. The tech industry welcomed the move, but it faced opposition from Democrats and some MAGA Republicans alike.

Some state complaints against preemption

The order drew threats of legal challenges. Colorado Attorney General Phil Weiser threatened to sue, and attorneys general from 22 states and the District of Columbia separately pushed back on Federal Communications Commission efforts to preempt state AI laws, two of those attorneys general being Republicans.

Policy analysts warned in April that federal preemption risks creating a regulatory ceiling rather than a baseline, weakening protections for workers and consumers.

The Obernolte-Trahan draft would give that preemption drive congressional backing, and potentially resolve the legal uncertainty clouding BEAD implementation. But it still faces daunting odds.

The administration failed to include preemption in two pieces of legislation in 2025. A provision in the One Big Beautiful Bill reconciliation package would have imposed a 10-year moratorium on state AI laws.

Sen. Marsha Blackburn, R-Tenn., axed the moratorium, forcing a vote that passed 99-1.

By pairing preemption with mandatory transparency and third-party audits, and limiting the freeze to three years rather than 10, the two lawmakers aim to make the trade-off politically workable.

Cybersecurity executive order on Tuesday

Whether the Trump administration backs the bill remains unclear. Earlier this week, Donald Trump signed a scaled-back AI cybersecurity order that retreated from stricter industry oversight, a sign of tension over how hard to push AI developers.

Trahan's involvement puts her at odds with House Democratic leaders building their own party-line AI agenda for the midterms. If they conclude she has conceded too much on preemption, they may keep their own framework on a separate track.

Worker and consumer advocates have argued that a preemption provision strips states of tools they are already using,arguments that brought down the reconciliation bill moratorium last year.

The bill's introduction as a discussion draft signals the authors want feedback before pushing to a formal vote. Congress breaks for recess in August.