Inside the Great American AI Act: Preemption, Transparency and a Fight Already Under Way

A bipartisan House proposal would move the core of AI governance from the states to Washington, and the fight over that shift began within hours of its release.

U.S. AI regulation has lived in the states. A bipartisan draft would move the core of it to Washington, and the argument over that shift began almost as soon as the text appeared.

On June 4, Reps. Jay Obernolte, a California Republican, and Lori Trahan, a Massachusetts Democrat, released a discussion draft titled the Great American Artificial Intelligence Act, law firm DLA Piper noted in an analysis. The text runs 269 pages and has not been formally introduced, Roll Call reported. It is a discussion draft, circulated to gather feedback from stakeholders, experts and the public before a bill is filed, the lawmakers said in a release.

The draft is bipartisan by design. Obernolte and Trahan published a joint op-ed in Bloomberg Law arguing that durable policy for a technology this consequential has to be written by both parties, DLA Piper reported. Reps. Scott Franklin, Suhas Subramanyam, Erin Houchin and Scott Peters joined in releasing it, according to the sponsors.

Preemption is the central fight

The most contested provision concerns who governs how AI systems are built. The draft would preempt state laws on the development of AI systems for three years, while leaving states free to regulate how those systems are used within their borders, Roll Call reported. A document from Trahan’s office said California’s AB 2013, which requires developers to post high-level summaries of their training data, would be preempted, along with part of SB 942 on content watermarking, according to Roll Call.

Supporters frame preemption as a fix for fragmentation. Their argument, made in the sponsors’ Bloomberg Law op-ed, is that AI systems built in one state are used in all 50, so protections that vary by zip code leave gaps, as DLA Piper reported. Rep. Houchin argued that a patchwork of 50 different state laws would make it harder for U.S. companies to compete with China while doing little for consumer protection, according to the sponsors’ release.

Critics see the same provision as the problem. The part that blocks states from legislating on how systems are built is the piece that matters most for safety, Broadband Breakfast reported from the draft’s opponents, and the House Democratic Commission on AI came out against the text within hours of its release, DLA Piper noted. That places the draft inside a larger argument about whether national uniformity or state experimentation better serves both safety and innovation.

What else the draft would do

Beyond preemption, the draft would require the most powerful frontier labs to open up their models, Broadband Breakfast reported. It would authorize $100 million per fiscal year for a Center for AI Standards and Innovation, call for accountability in government AI adoption, and formally codify the National Artificial Intelligence Research Resource, a shared pool of data and computing for AI research, FedScoop reported. The draft arrived two days after a White House executive order on AI that asked for early access to models for federal oversight, according to FedScoop.

Industry groups gave early, qualified support. The Business Software Alliance welcomed the draft as a serious starting point for advancing frontier-model safety and public trust, BSA said. The reception elsewhere was split, which is typical for a draft built to start a negotiation.

For the companies that build and deploy AI, the fight matters most as a question of cost and certainty. A single federal standard would replace the prospect of following rules that differ from one state to the next, a fragmentation that is expensive to navigate across a market of 50 states. That is the trade critics keep pressing: a national rulebook is easier to comply with, and it also strips states of authority over how the most powerful systems are built.

A discussion draft is an early marker, and this one faces organized opposition and a long path through committee before it could become law. Its value at this stage is as a statement of direction. The U.S. is weighing a federal AI framework built on two ideas: preempting state rules on how systems are developed and pushing the largest labs toward greater transparency. Both will be contested line by line, and the version that emerges, if one does, will shape how AI is governed in the world’s largest market.

MORE ON AI: