- The White House released its National Policy Framework for Artificial Intelligence on March 20, 2026, the most detailed federal AI legislative blueprint to date.
- The framework establishes six priorities including child safety, energy costs, intellectual property, and federal preemption of state AI laws.
- Brookings fellows Tom Wheeler and Bill Baer argue the framework “sidesteps the most important question in AI governance” by failing to establish any enforcement mechanism or oversight agency.
- The framework contains no provisions for mandatory incident reporting, frontier model safety testing, or restrictions on AI use in law enforcement.
What Happened
The White House released its National Policy Framework for Artificial Intelligence on March 20, 2026, outlining legislative recommendations intended to guide Congress as it considers federal AI legislation. The framework is not a binding document and does not create new legal obligations or direct agencies to take specific regulatory actions.
On March 31, 2026, Brookings Institution fellows Tom Wheeler and Bill Baer published an analysis arguing that the framework addresses symptoms rather than root causes of AI governance problems. Wheeler, a former FCC chairman, and Baer, a former assistant attorney general for antitrust, contend the framework “sidesteps the most important question in AI governance: who is in charge of those in charge?”
Why It Matters
The framework’s most consequential provision is federal preemption. Four states, Colorado, California, Utah, and Texas, have already enacted AI legislation. The White House explicitly advocates overriding state laws that the administration considers overly burdensome, arguing that a patchwork of 50 different regulatory regimes would hinder innovation. This positions the framework as both a regulatory proposal and a preemptive strike against state-level AI governance.
The Brookings analysis identifies a fundamental gap: the framework contains no institutional mechanism for oversight or enforcement. Wheeler and Baer write that “power does not regulate itself,” arguing that the framework delegates AI policy to Congress through aspirational language while avoiding the creation of any agency with actual authority to establish behavioral expectations or compliance mechanisms.
Technical Details
The framework establishes six priorities. First, protecting children from AI-generated harmful content, including age assurance mechanisms, limits on data collection, and parental oversight tools building on the Take It Down Act targeting deepfake abuse. Second, managing energy costs from AI data centers through streamlined permitting and safeguards to prevent increased consumer electricity costs. Third, preserving intellectual property rights in AI training. Fourth, preventing political censorship in AI systems. Fifth, educating Americans on AI literacy through apprenticeships and land grant institutions. Sixth, maintaining US global competitiveness.
The framework recommends against creating a new federal AI regulatory agency, instead directing existing agencies to adapt their oversight to cover AI within current jurisdictions. It also includes incentives to drive small business AI adoption and investment in workforce training and labor market analysis.
Who’s Affected
AI companies operating across multiple states would benefit most from federal preemption, which would replace a patchwork of state regulations with a single federal standard. Large technology companies including Google, Meta, Microsoft, and OpenAI have lobbied for federal preemption as a way to avoid compliance with dozens of different state-level requirements. States that have already passed AI legislation, particularly Colorado’s comprehensive AI Act and California’s proposed frontier model safety requirements, could see their laws overridden.
The absence of mandatory incident reporting and frontier model safety testing requirements means AI developers face no new compliance obligations from the framework itself. This contrasts sharply with the EU AI Act, which has established binding requirements for high-risk AI systems including mandatory conformity assessments and post-market monitoring. Companies operating in both the US and EU markets will continue to face asymmetric regulatory expectations.
Civil society organizations and consumer advocates have criticized the framework’s lack of enforcement provisions. Without an agency empowered to investigate complaints, impose penalties, or mandate disclosures, affected individuals have limited recourse when AI systems cause harm in areas such as hiring, lending, or insurance underwriting.
What’s Next
The framework is a set of recommendations, not law. Congress must draft, debate, and pass legislation to give any of its provisions legal force. Given the current congressional calendar and political dynamics, substantive AI legislation is unlikely before 2027 at the earliest. In the interim, state legislatures are likely to continue passing their own AI laws, creating exactly the regulatory patchwork the framework seeks to prevent. The question Wheeler and Baer pose, who is in charge of those in charge, remains unanswered.
Related Reading
- White House Releases Federal AI Framework, Urges Congress Against Heavy Regulation
- White House Releases National Policy Framework for AI, Setting Federal Guardrails
- White House Releases AI Legislative Framework Calling for Federal Preemption of State Laws
- California sets its own AI rules for state contractors, pushing back against federal policy
