The White House’s March 24, 2026 legislative recommendations on artificial intelligence (the National Policy Framework for Artificial Intelligence) mark a notable shift in U.S. AI policy. Building on Executive Order 14365, the Administration moves away from a largely hands-off approach toward a more centralized federal framework, albeit still non-binding and narrowly focused, particularly on child protection.
At the core of the very short Framework that is legally not binding are two interrelated developments: (1) an increased reliance on federal preemption to curb state-level AI regulation, and (2) the creation of an enforcement mechanism in the form of an AI Litigation Task Force.
From State Fragmentation to Federal Control
In recent years, U.S. states have enacted a growing patchwork of AI laws, often imposing detailed and divergent compliance obligations on businesses. The Framework explicitly targets this fragmentation. It envisions federal minimum standards, especially in sensitive areas such as child safety—and, where necessary, exclusive federal regulation.
This approach elevates preemption to a central policy tool. The Administration appears willing to displace state laws that are deemed overly burdensome to innovation or inconsistent with federal objectives, particularly in the run-up to midterm elections where targeted AI safeguards enjoy public support.
AI Litigation Task Force as Enforcement Mechanism
A key institutional innovation is the already proposed AI Litigation Task Force. Rather than serving a purely advisory role, the Task Force is designed as an active enforcement body with three primary functions:
- Identifying state AI laws that impose undue burdens on innovation;
- Assessing their constitutionality, particularly under the Constitution’s Commerce Clause; and
- Initiating or supporting litigation to invalidate conflicting state measures.
This signals a more assertive federal posture, using litigation strategically to achieve regulatory uniformity with an uncertain time frame.
Federal-State Tensions and Practical Constraints
The Framework’s centralizing ambition is likely to face resistance. States have long functioned as “laboratories of democracy,” particularly in emerging technology regulation. The Litigation Task Force may therefore become a focal point of constitutional disputes over the limits of federal authority.
Complicating matters, a key prerequisite for enforcement—the Department of Commerce’s report identifying “problematic” state laws—remains outstanding past its deadline. This delay introduces uncertainty for both regulators and industry and may reflect political sensitivities in an election cycle.
International Positioning and Limited Substantive Guidance
The Framework also underscores the U.S. goal of shaping global AI governance, emphasizing competitiveness and alignment with American values. At the same time, it offers limited substantive guidance in key areas. Notably, on AI and copyright, the Administration declines to take a legislative position, effectively deferring to the courts—particularly on the scope of fair use in AI training. This hands-off approach prolongs legal uncertainty for AI developers.
Parallel Legislative Developments: The “Trump America Act”
Separately, policymakers continue to advance alternative federal AI frameworks. For example, Senator Marsha Blackburn has introduced a Republican-led proposal referred to as the “Trump America Act,” which outlines a national AI policy emphasizing innovation, limited regulation, and U.S. technological leadership. While politically aligned with broader deregulatory goals, its prospects for enactment appear low in the current divided Congress, particularly given competing priorities and the proximity of elections.
Outlook
The White House’s approach reflects a broader recalibration of U.S. AI policy: regulation is no longer viewed as inherently detrimental, but as necessary—provided it is federally coordinated and enforcement-backed. The combination of policy guidance and litigation tools represents a shift toward a more centralized and, if necessary, adversarial federalism. With this approach the Framework seems to move away somewhat from earlier statements of the White House that no AI regulation is necessary at all.
One reason for this chance is that Republican candidates in the current election cycle are increasingly campaigning on platforms that emphasize limited, sector-specific AI regulation, prioritizing innovation and avoiding broad federal constraints on emerging technologies. At the same time, polled public opinion in the United States continues to support targeted AI safeguards—particularly in areas such as child protection and consumer safety—rather than a wholly unregulated approach.
At the same time, state-level legislative activity continues unabated, underscoring ongoing regulatory competition. At the state level, Washington State recently enacted legislation regulating “companion chatbots,” imposing transparency, safety, and conduct requirements on AI systems that engage in human-like interactions with users, especially minors. Whether the federal government can effectively impose uniformity will depend on congressional action, judicial interpretation, and political dynamics following the elections.
