Great American AI Act: Congress Moves to Freeze State AI Laws

Two House lawmakers dropped the most comprehensive federal AI bill yet on Thursday, June 4. Reps. Jay Obernolte (R-CA) and Lori Trahan (D-MA) released a 269-page discussion draft of the Great American Artificial Intelligence Act — and its centerpiece is a three-year freeze on state AI laws.

The draft would preempt state rules "specifically regulating the development" of AI models for three years, then sunset. It's the clearest sign yet that Congress wants a single national rulebook for frontier AI instead of a patchwork of state laws.

Why this fight is happening now

States have moved faster than Washington on AI, and that's the problem the bill is trying to solve. Colorado's AI Act — the first comprehensive state AI law in the country — is scheduled to take effect June 30, just weeks away. California has passed its own measures. Industry has spent the past year warning that 50 different state regimes would be impossible to comply with.

The timing isn't a coincidence. The White House issued an executive order in December 2025 specifically targeting Colorado's law, and this draft reads as the legislative follow-through. Freeze the states, build one federal framework, and give companies a single target.

Why this matters: where AI gets regulated determines who has leverage. State capitals have been more willing than Congress to impose binding anti-discrimination and safety rules. Moving the fight to Washington changes the odds.

What the preemption actually covers

The preemption is narrower than the headlines suggest, and the distinction matters. It applies to laws regulating the development of AI models — the training and building of frontier systems. It does not touch laws about how AI is used or deployed.

States could still pass laws of "general applicability" — rules that apply to AI the same way they apply to everything else — and could regulate models after they've been deployed. So a state consumer-protection or anti-fraud law would likely survive; a state law dictating how a frontier model must be built would not.

That carve-out is where the legal fights will live. "General applicability" is vague enough to litigate for years, and critics argue the line between developing and deploying a model is blurry in practice.

What the bill requires from big AI companies

The draft isn't only preemption — it imposes real federal obligations on the largest players, defined as companies with more than $500 million in annual gross revenue.

Those companies would have to publish public Frontier AI Frameworks explaining how they govern their most capable models. They'd have to report critical safety incidents to the federal government, and allow auditors to verify their cybersecurity mitigation plans.

The bill also funds a Center for AI Standards and Innovation inside the Commerce Department at $100 million a year. And it reaches beyond model governance: it adds criminal penalties for using AI to impersonate government officials, directs the Census Bureau to add AI-usage questions to federal surveys, and extends the Cybersecurity Information Sharing Act of 2015 through 2035.

Who this affects

For frontier labs, this is a trade. They'd accept new federal transparency and incident-reporting duties in exchange for escaping a thicket of state rules — a deal most large AI companies have signaled they'd take.

For states, it's a loss of authority over the most consequential category of AI. Colorado's law, which includes actual anti-discrimination requirements in employment, lending, housing, and healthcare, could be frozen before it ever bites — if the bill becomes law.

For enterprises deploying AI, the near-term reality is uncertainty. If the Act passes, the compliance target shifts to Washington. If it stalls, Colorado's rules take effect June 30 and companies serving Colorado residents need frameworks in place now.

What insiders are saying

The reaction split along predictable lines. Tech industry groups including the Information Technology Industry Council (ITI) and NetChoice praised the draft. Labor unions rejected it outright: the AFL-CIO, AFT, and Association of Flight Attendants called it "a giveaway to the AI industry."

Even some safety advocates who like parts of the bill balked at the preemption. Brendan Steinhauser, CEO of the Alliance for Secure AI, praised its bipartisan focus on catastrophic risk and loss of control but opposed freezing state laws. The critique cuts across ideology: preempting state protections without equally strong federal ones, opponents argue, leaves a gap.

What's next

This is a discussion draft, not a bill that's been formally introduced — an important distinction. The authors released it to collect feedback from experts, companies, and the public before it moves. The legislative road ahead is long, and preemption has tanked AI provisions before.

Watch two clocks. The first is Colorado's June 30 effective date, which arrives long before Congress could act. The second is whether the White House, which has been skeptical of binding requirements on companies, decides to back the framework. Its support — or silence — will shape what survives.

Bottom line

The Great American AI Act is the most serious attempt yet to put one federal rulebook over AI, and its three-year state freeze is the provision everyone will fight over. If you run AI compliance, don't pause your state-law preparation — this is a draft with a hard road ahead, and Colorado's deadline is real and close. Track it, but don't bet your roadmap on it passing.