A bipartisan council of US governors asked Washington on 22 November to set rules for the most advanced AI models, while keeping states in charge of most protections for people, the National Governors Association said. Bloomberg reported the plan the same day.
By our count of the NGA’s statement of principles, 14 of its 24 principles are addressed to states and 5 to the federal government. The other 5 apply to both.
The council is led by Maryland Gov. Wes Moore and North Carolina Gov. Josh Stein, both Democrats, and Indiana Gov. Mike Braun and Utah Gov. Spencer Cox, both Republicans.
Business Pill · The Cost of Checking
An evergreen Business Engineer explainer. It teaches the general idea only and does not discuss this case.
The key insight: By our count, the NGA governors council’s AI statement addresses 14 of its 24 principles to states and 5 to Washington, asking Congress for frontier-model rules with independent audits and incident reporting while keeping states as a primary accountability mechanism for AI harms to people.
What the Governors Ask Washington
The first federal ask is a “risk-based framework” for frontier models, with standards for “the most advanced, general-purpose AI models whose risks and reach are national in scale.”
Under it, the models “should be evaluated or audited by independent experts,” and developers should report “security breaches and significant safety incidents to federal authorities and affected states,” the document says.
The other federal asks cover national security and critical systems, international cooperation, and American AI leadership. The fifth is titled “Approach preemption carefully.”

Preemption Is the Fault Line
“Any attempt to preempt state laws should be approached cautiously and include meaningful input from a representative cross-section of states,” the principles say.
Preemption means a federal law overriding state laws. The governors say they will judge any such attempt against the NGA’s standing “Principles of Federalism” policy.
The question is live. Wisconsin Gov. Tony Evers wrote in April that a national AI policy framework published on 20 March “would preempt state AI laws,” in a letter to his state’s congressional delegation. We covered an earlier federal freeze proposal here.
What Stays With the States
The 14 state principles open with consumer harm: “States should serve as a primary accountability mechanism for AI harms affecting people as AI is deployed.”
They also cover children and chatbots, scams aimed at older adults, AI in hiring, lending and housing decisions, public-sector use, schools, health care, privacy and cybersecurity.
By our grouping, 6 of the 14 are about protecting people from harm, 4 about workers, schools and talent, 3 about using AI in government and health, and 1 about states coordinating.

No Threshold Yet
The document does not define “the most advanced” models with a compute or revenue cutoff. It asks for “clear, technically informed definitions” with regular review instead.
California already uses a 10^26-operation compute threshold in its AI safety law, SB 53. Its own report on 18 November said relying only on that cutoff has left out some dangerous incidents, as we reported.
The NGA principles also set no deadline for reporting safety incidents. California’s report proposed cutting its main deadline from 15 days to 48 hours, according to the state’s report.
What the Governors Said
“We agree that humans must stay in control, and we must protect our people – especially workers, consumers and our kids,” Moore said in the NGA release.
“We don’t have to choose between innovation and safety,” Cox said. NGA chief executive Brandon Tatum said “no federal AI policy will be effective without governors’ input.”
The Structural Read
As we read it, the split is a bid to shape any federal law before it is written: Washington gets the frontier models, states keep the harms people feel day to day.
As we read it, the missing threshold matters, because who counts as a frontier developer decides how many companies the federal rules would reach.
NGA Statement of Principles, Nov 2026
“Any attempt to preempt state laws should be approached cautiously and include meaningful input from a representative cross-section of states.”
Three Implications
FOR AI LABS The governors ask that frontier models be audited by independent experts and that developers report significant safety incidents to federal and state authorities.
FOR CONGRESS The principles say any preemption of state laws should be approached cautiously, with input from a cross-section of states.
FOR STATES The 14 state principles say states should serve as a primary accountability mechanism for AI harms to people, including consumers, workers and children.
What Is Not Established
The release names the four council leaders but does not list every governor who backs the principles. It says NGA will engage governors’ offices from all states and territories next.
The principles are recommendations, not a bill. They are meant to inform “the incoming Congress,” the NGA said, and carry no legal force on their own.
The Bottom Line
The governors’ plan gives Washington the frontier-model rules and keeps most protections for people with the states, 14 principles to 5 by our count. What to watch is whether the incoming Congress writes a federal framework, and how far it preempts state laws.
97,000+ executives read Business Engineer for the AI strategy frameworks cited by ChatGPT, Claude, and Perplexity.
A note on sourcing. The principles, quotes and council leaders come from the National Governors Association’s release and its statement of principles, both dated 22 November 2026. Bloomberg’s report is cited from its headline and summary; we did not read its full text. The March framework is as described in Gov. Evers’s April letter. California’s threshold and deadlines are from the state’s 18 November report, as we reported. The 14-of-24 count, the 58% share and our grouping of the state principles are our arithmetic.
Sources: National Governors Association, ‘Governors Call for AI Action’, press release, 22 Nov 2026 · NGA Governors Council to Secure American AI Dominance, ‘Federal and State Roles in AI Regulation: Statement of Principles’, Nov 2026 · Bloomberg, ‘Governors Call for Federal AI Rules Without Sidelining States’, 22 Nov 2026 · Office of Wisconsin Gov. Tony Evers, letter to Wisconsin’s congressional delegation on federal AI preemption, April 2026 · California Government Operations Agency, ‘AI Safety in 2026 and Beyond’, 18 Nov 2026









