On March 20, the White House released a national legislative framework urging Congress to preempt state‑level artificial intelligence (AI) laws that it views as overly burdensome and inconsistent, signaling a push to centralize AI regulation at the federal level. The framework reflects the administration’s concern that state AI laws could hinder innovation, investment, and the development of a consistent national AI strategy. According to the White House, a single federal standard is needed to avoid conflicting compliance obligations and to support the U.S. competitiveness in AI.
This framework document follows and is guided in concept by Executive Order 14365, entitled Ensuring a National Policy Framework for Artificial Intelligence, which was issued on December 11, 2025. That executive order called for subsequent actions to address various state laws that create what the President describes as a “patchwork” of regulatory regimes concerning AI that make compliance difficult in the marketplace
Prior federal attempts to address AI regulation have been limited, with the most recent legislative attempt being efforts by some Republicans in Congress trying, unsuccessfully, to include a provision in the 2026 National Defense Authorization Act that would have imposed a blanket moratorium on the states from regulating AI for ten years. That proposed legislation took a blanket approach to the risk of possible inconsistent state laws, without making sufficient exceptions for a number of practical concerns that many interest groups felt were proper roles for states, such as child safety, AI data center infrastructure, and state government use of AI. In particular, it proposed preemption without creating any viable federal legislation or regulation concerning the use and deployment of AI. The pushback from many states and interest groups on both sides of the aisle likely doomed that legislation, given its broad scope and its lack of alternatives.
The debate over AI preemption echoes long‑running issues in data privacy regulation. To date, Congress has refrained from enacting a comprehensive federal privacy law, and states have largely been left to adopt generally applicable privacy laws of their own. That approach, however, has produced mixed results. State laws have allowed regulatory experimentation, but they have also created overlapping and sometimes conflicting compliance requirements. In the privacy law realm, federal legislation is mainly sectoral-specific, i.e., legislation and corresponding regulations providing privacy rights and compliance regimes for industries such as health care, financial lending, and online child protection rules.
But it has been at the state level where comprehensive, non-sectoral laws and regulations designed to protect individuals’ privacy regardless of the industry or business collecting and using their data, especially in online settings, have been implemented. The various state efforts have led to a myriad of different rules, ranging from differences in notification of intended data collection, use, and sharing to rules providing for whether individuals or government officials have the right to file suit to enforce those laws. The costs to businesses in terms of resources and clarity is one of the considerations that the proponents of preemption in the AI space are trying to avoid, and thus prevent unwarranted delays in AI development and deployment in the U.S.
The White House’s AI framework appears aimed at avoiding a similar outcome by advancing federal preemption earlier in the development of AI regulation. While the newly-released framework supports broad federal preemption, the administration apparently took note of the deficiencies of its congressional allies’ approach and identified key areas where state authority would remain intact. Citing the importance of maintaining proper concepts of federalism by preserving appropriate state authority for general police powers, zoning regulation, and state governments’ own use of AI, the White House further emphasized that a national AI standard should not displace state laws addressing fraud, consumer protection, or child safety. The framework cautions that federal preemption should be limited to areas better suited for federal oversight or where state laws conflict with national AI objectives.
The framework addresses the use of copyrighted materials to train AI models, expressing support for allowing courts to resolve whether such practices violate copyright law, rather than congressional intervention. It also suggests that Congress consider mechanisms, such as licensing frameworks or collective rights organizations, that would allow rights holders to negotiate compensation from AI providers without incurring antitrust liability. The framework further proposes exploring federal protections against unauthorized commercial use of AI‑generated replicas of individuals’ likenesses or voices, with exceptions for news reporting and satire. AI infrastructure and energy costs are also key themes. The White House urged Congress to ensure that residential ratepayers do not bear increased electricity costs associated with new AI data centers, while streamlining federal permitting to accelerate AI infrastructure development and support grid reliability.
Whether Congress will act remains uncertain. Prior efforts to limit state AI regulation have stalled amid resistance from states and lawmakers. The framework, however, does signal an effort to assert federal primacy over AI regulation, while leaving key details to future legislation and judicial interpretation. The comparison to federal privacy law underscores both the challenges of regulatory fragmentation and the difficulty of achieving federal consensus. Entities developing or deploying AI should continue to monitor congressional activity, litigation, and evolving state laws shaping the regulatory landscape.
