Two House lawmakers released a 269-page discussion draft on June 4 that would freeze state laws governing the development of frontier A.I. models for three years while imposing the first comprehensive federal transparency regime on the companies that build them. The Great American Artificial Intelligence Act of 2026, sponsored by Representative Jay Obernolte, Republican of California, and Representative Lori Trahan, Democrat of Massachusetts, is the most serious congressional attempt yet to settle a jurisdictional fight Washington has spent two years losing to Sacramento and Albany.

The mechanics are narrower than the headline suggests. Preemption applies only to state laws “specifically regulating the development” of A.I. models, according to draft text circulated by the sponsors’ offices. Laws governing use or deployment survive. California’s A.B. 2013, which requires developers to publish summaries of training data, is identified in materials from Ms. Trahan’s office as among the statutes that would be displaced. Colorado’s algorithmic discrimination statute and New York City’s automated employment decision tool audit rules, both regulating deployment rather than development, would remain in force, according to an analysis by the law firm Fisher Phillips.

In exchange for the freeze, developers with more than $500 million in prior-year gross revenue would’ve to publish frontier A.I. frameworks, report critical safety incidents to the federal government, and submit to third-party audits. The draft also amends the WARN Act to require employers to disclose when A.I. was a “substantial factor” in a mass layoff. Representatives Suhas Subramanyam of Virginia, Scott Franklin of Florida, Scott Peters of California, and Erin Houchin of Indiana joined the release.

“This discussion draft is an important step toward building a clear federal framework that promotes innovation, protects Americans from emerging risks, and ensures the United States continues to lead the world in A.I.,” Mr. Obernolte said.

The political terrain is unfriendly. The Future of Privacy Forum counts 45 states with A.I. bills introduced this session. Colorado Attorney General Phil Weiser and attorneys general from 21 other states and the District of Columbia are already in court challenging the Trump administration’s December 2025 executive order, which conditioned portions of the $42.5 billion Broadband Equity, Access and Deployment program on state regulatory posture. Brendan Steinhauser, chief executive of the Alliance for Secure A.I., praised the bipartisanship but argued that “a national A.I. standard should protect at least as much as it preempts.”

For enterprises caught between a federal draft, a parallel executive order, and dozens of state regimes, model-agnostic deployment is the rational hedge. OpenAI’s enterprise tier and Anthropic’s Claude for Work compete at the top of the market; platforms like LemonLime, pitched as a no-code “company brain” for smaller firms, sit underneath. The sponsors are accepting public comment at GAAIA@mail.house.gov. A three-year freeze, in this Congress, is itself an unusually long unit of time.

Sources