White House Drops AI Legislative Blueprint — Pushes to Preempt State Laws

The White House released its National Policy Framework for Artificial Intelligence on March 20, a four-page legislative blueprint that tells Congress exactly what the administration wants: light federal regulation, broad preemption of state AI laws, and no open-ended liability for AI developers.

The document was written by Michael Kratsios, White House science and technology adviser, and David Sacks, Special Adviser for AI and Crypto, fulfilling a directive from President Trump’s December executive order on AI.

The seven policy areas

The framework lays out Congress’s homework across seven categories: child safety, community effects from AI infrastructure, copyright, indirect government censorship (what the document calls “jawboning”), federal regulation scope, jobs and workforce training, and state law preemption.

State preemption is the headline item. The framework explicitly calls for blocking states from regulating AI model development or imposing liability on AI developers for unlawful conduct by third parties using their systems. States would retain authority over generally applicable laws, data center zoning, and procurement decisions — but not AI-specific regulation.

Why this matters

Right now, at least 17 states have proposed or passed AI-specific legislation. California’s SB 1047, though vetoed in 2024, established the template: require safety evaluations before deploying frontier models. Colorado enacted AI-specific consumer protection rules in 2024. Illinois and Texas have bills in committee.

The federal framework aims to create a single regulatory environment instead of a patchwork. For AI companies, this is a clear win — compliance with 50 different state regimes is expensive and unpredictable. For state legislators who view AI regulation as urgent, it’s a direct challenge to their authority.

Industry and critic reactions

NetChoice’s Patrick Hedger praised the framework, saying it demonstrates “what is at stake and what it will take to win” the AI competition with China. Daniel Castro at the Center for Data Innovation noted the framework avoids “alarmism” about job displacement.

Brad Carson at Americans for Responsible Innovation called it “another chance for tech companies to launch harmful products with no accountability.” The criticism highlights the core tension: preemption removes both burdensome regulation and protective regulation simultaneously.

The copyright question

The framework addresses AI and copyright but stops short of specifics. It acknowledges the tension between training data usage and creator rights without proposing a resolution. With the New York Times v. OpenAI case still in litigation and the Copyright Office working on its own guidance, Congress is being asked to legislate in an area where the courts haven’t finished deciding the existing rules.

What happens next

This is a blueprint, not legislation. Congress has to write actual bills, and the preemption provisions will face opposition from state attorneys general and consumer advocacy groups. The framework’s practical impact depends on whether any of these recommendations survive the legislative process — and in an election cycle, AI regulation may not be the priority either party wants to own.