This Fall, Congress Should Finish the Job on a National AI Framework

America’s artificial intelligence policy debate has moved quickly over the past year, but one fundamental question remains unresolved: will the United States have a national framework for AI, or will national policy increasingly be written state by state? Congress should make answering that question one of its top priorities when it returns this fall.

The case for federal leadership has only grown stronger. AI is no longer an emerging technology used by a relatively small group of companies and researchers. Americans are using it every day to start and grow businesses, improve productivity, learn new skills, obtain health information, manage household finances, and solve ordinary problems. AI is also becoming more important to medicine, manufacturing, transportation, national security, and scientific research. A technology that operates across the entire national economy should not be governed by an increasingly fragmented set of state rules.

The Trump administration has called for a federal AI policy framework and has argued that Congress should preempt state AI laws that impose undue burdens while preserving traditional state authority in areas such as fraud, consumer protection, child protection, zoning, and state government operations. Congress now needs to turn that principle into legislation.

A national framework does not require Washington to regulate every use of AI or strip states of their traditional powers. It means identifying the issues that are inherently national and setting national rules for them. The development of AI models is an obvious example. An AI company does not build one model for California, another for Texas, another for Colorado, and another for Virginia. Models and services operate across state lines instantly. Forcing developers, startups, and small businesses to navigate substantially different rules governing the same technology in dozens of jurisdictions would create enormous compliance costs and discourage innovation.

There is also a broader strategic problem. The United States is competing globally to lead in AI. If American companies face a growing maze of overlapping and contradictory state rules while foreign competitors operate under more coherent national systems, we are creating a disadvantage for ourselves. Congress should not allow the rules governing one of the most important technologies in the world to emerge piecemeal through a collection of unrelated state statutes.

A national framework would also give businesses and consumers something they both need: clarity. Companies should know the rules before investing billions of dollars, hiring workers, building infrastructure, or launching products. Americans should be able to expect basic protections that do not fundamentally change when they cross a state line. Lawmakers should identify specific harms and address them directly rather than allowing broad restrictions on AI development to become substitutes for targeted policy.

Congress has already shown that it can address specific technology-related harms without trying to regulate AI as a single undifferentiated category. That approach should continue. Congress should address real risks where they exist, preserve appropriate state authority, and establish national rules where national consistency is necessary.

Waiting will only make the problem harder. Every legislative session gives states another opportunity to create new requirements. Every new requirement makes the eventual task of building a coherent national framework more complicated. Once companies have redesigned products, compliance systems, and business models around dozens of different regimes, unwinding that structure becomes more difficult and more expensive.

Congress should not treat a national AI framework as an issue that was debated earlier this year and can now be set aside. It remains unfinished business. The goal should be a durable division of responsibility: national rules for inherently national questions, continued state authority over traditional state responsibilities, and enough regulatory certainty for American innovators to build. That is a much better approach than allowing fifty different policy experiments to slowly become the de facto national AI regime.

AI policy will help determine where investment goes, which companies succeed, what technologies Americans can use, and which country leads the next era of technological development. Congress should not leave those questions to be answered accidentally, state by state. When lawmakers return this fall, establishing a meaningful national AI framework, including strong and carefully drawn federal preemption, should be near the top of the agenda.

Neil Siefring is a senior fellow at Alliance for the Future.