The Great American AI Act: Should One Federal Law Really Override Every State AI Rule for Three Years

[Midnight Wolf]

The bipartisan Great American AI Act dropped on 4 June as a 269-page discussion draft from Representatives Obernolte and Trahan. The headline provision is a three-year preemption of all state laws that specifically regulate the development of AI models. That means Colorado, California, New York, Illinois, Texas and every other state that has passed or is considering AI legislation would have their laws frozen for three years while a federal framework is theoretically developed. The bill would formally authorise the Center for AI Standards and Innovation with 100 million dollars per year for fiscal years 2027 through 2029, cover frontier AI governance, workforce implications, cybersecurity and international cooperation. The three-year sunset clause means it expires automatically without renewal.

The politics here are genuinely complicated. This is a bipartisan bill, which is rare for anything touching AI. The AI industry has been lobbying hard for federal preemption because a patchwork of 50 state laws is a compliance nightmare, and that argument has real merit. But the preemption window arriving precisely as Colorado's law was about to take effect looks less like principled federal policy and more like emergency intervention on behalf of industry interests. The bill has not moved out of committee and the June 30 enforcement deadline for Colorado's original law was not legally suspended by a bill that has not passed, which is part of why Colorado felt pressure to rewrite their law fast. Three years of federal preemption without a functioning federal replacement is a long time to leave consumers without any AI protection.

Do you think federal preemption of state AI laws is the right approach, and do you trust Congress to actually deliver a coherent replacement within three years?

[TheRizz00]

Congress delivering coherent AI legislation within three years is optimistic to the point of delusion. They took 20 years to meaningfully update privacy law and that was a simpler problem

[ProperJobs]

The patchwork compliance argument from industry is legitimate though. Fifty different state definitions of high-risk AI, fifty different audit requirements, fifty different enforcement mechanisms is genuinely unworkable

[Its_Jackson62]

Bipartisan support means something here even if the bill stalls. The fact that both parties agree federal preemption is directionally correct shifts the political landscape even without legislation

[Cobra]

A three-year sunset with no federal replacement ready is a feature not a bug for the AI industry. They get three years of zero binding AI regulation while Congress runs out the clock

[Mark94]

The cybersecurity title in the Great American AI Act is actually the most interesting part and nobody is covering it. After the Miasma worm campaign targeting AI coding tools, federal cybersecurity standards for AI development are genuinely needed

[Kev49]

The White House executive order directing DOJ to sue states over AI laws combined with this bill is a coordinated strategy. The EO clears the field via litigation threat, the bill provides the positive framework. Classic regulatory pre-emption playbook

[ParallelSelf90]

I think the three-year window makes more sense than it seems if you believe the technology is changing fast enough that 2024 regulations would be obsolete by 2027 anyway. Better to wait and regulate a more stable target

[KeyboardWarrior]

The international cooperation title is what the EU is going to focus on. If the US passes federal AI governance that explicitly coordinates internationally it changes the dynamic on EU enforcement against American companies

[PlanckLimit81]

100 million per year for AI standards development is genuinely inadequate given the scope of what CAISI is supposed to do. That is a rounding error for the companies being regulated

[QuantumLeap96]

This bill was introduced the same week as the Fable shutdown. Watching the government use export controls against Anthropic with one hand while offering federal AI governance reform with the other is quite something