A bipartisan pair of U.S. House lawmakers released draft legislation on Thursday aimed at stopping states from regulating the development of artificial intelligence models.
Democrat Lori Trahan and Republican Jay Obernolte published the draft, which would bar states from adopting laws targeting artificial intelligence model development.
The language means states would lose the authority to require testing before AI companies release models to the public, according to the draft.
The proposal draws a line between AI model development and AI use. States would still regulate how organizations apply AI technology in specific settings.
Trahan and Obernolte said the measure seeks to create a national framework to protect Americans, support innovation, and keep the U.S. ahead in shaping AI technology.
Obernolte said the lawmakers released the draft to collect views from stakeholders, technical experts, and the public before they formally introduce the bill. The process gives Congress another run at AI rules after years of stalled attempts.
Federal lawmakers have struggled to pass legislation addressing AI safety concerns. That vacuum has pushed many states to draft their own rules, including measures tied to testing, consumer protection, civil rights, and transparency.
Public Citizen criticized the bill and said it would leave oversight largely with a federal government which has repeatedly failed to pass meaningful AI protections. The consumer advocacy group has argued for urgent state-level AI regulation, given the lack of federal laws.
The group also said the draft does not address algorithmic discrimination, housing discrimination, employment discrimination, consumer fraud, youth mental health harms, AI companions, deepfake exploitation, or market concentration. A long list, and not a small one.
Technology industry groups read the proposal differently.
The Information Technology Industry Council, which represents major tech firms, praised the draft. The group said Congress must set a national standard for responsible AI development, deployment, and adoption across the American economy.
The White House urged lawmakers in March to pass legislation pre-empting state AI rules. Earlier this week, President Donald Trump signed an order asking leading AI developers to voluntarily submit their most capable models for government cybersecurity tests before public release.
Under the order, U.S. agencies would receive up to 30 days to test those models before companies release them to organizations outside the federal government. The order also directs agencies to strengthen cyber defense across government systems.
Trump said in December he would withhold federal broadband funding from states whose AI laws his administration judges as holding back American dominance in the technology. That threat added pressure to a debate already split between state control and federal pre-emption.
The AI industry has become a major profit engine for the technology sector in recent years. Demand for AI chips helped Nvidia become the world’s largest company, while Amazon, Meta Platforms, Alphabet, and Microsoft continue to pour bn of dollars into AI infrastructure, models, and services.
For insurers and technology risk specialists, the draft raises familiar governance questions. A single national rulebook reduces compliance fragmentation for AI developers.
It also narrows the role of state regulators, which often move faster than Congress on consumer harm, discrimination, and emerging liability.
We think the bill’s main conflict sits there. Federal consistency helps large AI firms scale products across the U.S. market. State limits, according to consumer advocates, remove one of the few active oversight channels still available while Congress debates.
