Reps. Jay Obernolte (R-Calif.) and Lori Trahan (D-Mass.) released a 269-page discussion draft of the Great American Artificial Intelligence Act on June 4, proposing a three-year federal preemption of state AI development rules paired with a semi-annual third-party audit regime for any lab with more than $500 million in annual revenue. It’s the most substantive bipartisan attempt yet to do what the executive branch has been doing piecemeal: convert the live patchwork of state AI law into a single federal regime, on terms the largest labs can plan around.
The draft, which carries four additional co-sponsors alongside Obernolte and Trahan, defines a class of “large frontier developers” that effectively maps to OpenAI, Anthropic, xAI, and Google DeepMind. Those companies would be required to retain licensed independent auditors every six months, with full access to company records, personnel, and systems. Civil penalties run up to $1 million per violation per day. The bill funds a new Center for AI Standards and Innovation inside the Commerce Department at $300 million over three years. The auditing regime and the preemption clause both sunset after three years.
The bill defines catastrophic risk as incidents involving the death or serious injury of more than 50 people, more than $1 billion in property damage, WMD or cyber events, or “harmful autonomous action without meaningful human oversight.” AI-specific development transparency laws in California, New York, and Illinois would effectively be frozen.
The politics here are doing more work than the text. In December 2025, an executive order directed DOJ to challenge state AI laws on constitutional grounds; in April, DOJ moved against Colorado’s algorithmic discrimination statute. Days before the draft dropped, President Trump signed an order establishing voluntary federal reviews of new frontier models. Obernolte and Trahan aren’t legislating into a vacuum. They’re offering Congress a way to ratify a preemption posture the executive branch is already pursuing in court, with audit obligations attached as the price of admission.
Reactions split along the fault line you’d expect. The Information Technology Industry Council called it “an important step toward building a clear federal framework.” Brad Carson of Americans for Responsible Innovation said it “takes the current floor on state AI legislation and turns it into a federal ceiling,” and ARI launched ads in Massachusetts pressing Trahan to drop the preemption. Brendan Steinhauser of the Alliance for Secure AI said it “falls short of the framework necessary to keep Americans safe.” Trahan, defending the structure, said it “establishes real accountability for the most powerful frontier systems.” Rep. Erin Houchin (R-Ind.) framed the alternative as “a patchwork of fifty different state laws.”
What’s striking isn’t the trade itself, it’s how cleanly it’s been priced. Frontier labs get three years of regulatory legibility; safety advocates get an audit regime they don’t trust; states lose the instrument they’ve spent two years building. Each side knows exactly what it’s giving up, which is usually the sign a deal is closer than the press releases suggest.
Sources
- https://obernolte.house.gov/media/press-releases/obernolte-trahan-release-discussion-draft-great-american-ai-act
- https://rollcall.com/2026/06/04/bipartisan-ai-draft-proposes-three-year-preemption-of-state-laws/
- https://news.bgov.com/bloomberg-government-news/house-ai-framework-addresses-state-rules-promotes-standards
- https://broadbandbreakfast.com/ai-preemption-battle-lands-in-congress-with-substantive-discussion-draft/
- https://www.techtimes.com/articles/317903/20260606/federal-ai-regulation-bill-freezes-state-consumer-protections-three-years-sparks-revolt.htm