Summary
On June 5, 2026, Congress introduced the “Great American Artificial Intelligence Act,” a comprehensive bipartisan bill spanning 269 pages that represents the most ambitious federal attempt to regulate artificial intelligence to date. The legislation proposes a three-year moratorium on state-level AI laws, mandatory risk management programs for developers of frontier AI models, and the formalization of a federal AI standards body.
The bill would also establish workforce monitoring programs to track AI’s impact on employment, strengthen cybersecurity requirements for AI systems, and significantly boost federal AI research and development funding. The legislation arrives on the heels of President Trump’s recent executive order establishing voluntary federal agency reviews for frontier AI models before deployment in sensitive government contexts.
The timing is significant: Colorado’s SB 205, which requires risk management programs and impact assessments for “high-risk” AI systems, is set to take effect on June 30. Meanwhile, Illinois has sent five AI bills to its governor (including a frontier model safety bill), California has advanced 30 AI bills through committees, and Vermont has banned “therapy bots.” The federal bill’s three-year state preemption clause would freeze this wave of state-level regulation in its tracks.
Sources
Commentary
This bill is essentially the tech industry’s dream scenario wrapped in bipartisan packaging: a federal framework that simultaneously preempts the patchwork of state laws that companies have been dreading. The three-year freeze on state AI legislation is the most consequential provision — it would effectively kill Colorado’s SB 205 before it takes effect and halt momentum in California, Illinois, and other states that have been moving aggressively on AI regulation.
Whether this is good policy depends on your perspective. A unified federal standard would eliminate compliance chaos for companies operating nationally. But the “voluntary” nature of frontier model reviews (mirroring Trump’s executive order) and the state preemption combined with still-developing federal standards risks creating a regulatory vacuum where neither level of government is effectively overseeing AI deployment. The bill’s 269 pages will need serious scrutiny — the devil is always in the definitions of “frontier AI” and “high-risk systems” that determine who actually has to comply with what.