Who Writes the Rules: The White House AI Framework and What It Leaves Out

On March 20, 2026, the Trump administration released its National Policy Framework for Artificial Intelligence — a set of legislative recommendations to Congress that the White House describes as addressing "the most pressing policy topics that AI presents." The document organizes these topics into six objectives.

Five of them are about what the government can do for the AI industry. One is about what the AI industry might eventually do for workers.

That sentence is not rhetorical compression. It is a description of the document.

What the framework does

The first five objectives are, in order: protecting children, safeguarding communities (with emphasis on data center permitting and energy infrastructure), respecting intellectual property rights, preventing censorship, and enabling innovation. These are written with varying degrees of specificity. The preemption provision — which appears across multiple sections but is most explicit near the document's close — states that "a patchwork of conflicting state laws would undermine American innovation and our ability to lead in the global AI race," and calls on Congress to establish a unified national standard preempting state AI regulation.

The preemption is the operative policy. California, Colorado, and a dozen other states have passed or proposed AI-related legislation that includes worker protection provisions — disclosure requirements for AI-assisted hiring decisions, algorithmic discrimination safeguards, transparency mandates for AI in employment contexts. Under the federal preemption framework the White House is proposing, those state laws would not survive. The federal standard would replace them.

The federal standard includes no worker protection provisions.

The sixth objective

The framework's sixth and final objective is titled "Educating Americans and Developing an AI-Ready Workforce." It is the shortest section. Its operative language calls on Congress to "further workforce development and skills training programs, expanding opportunities across sectors and creating new jobs in an AI-powered economy."

Note the verb: "encouraging Congress." Not mandating. Not funding. Not establishing. Encouraging.

The previous administration's executive order on AI — EO 14110, issued October 30, 2023 — included workforce provisions with no enforcement mechanism. That order was rescinded January 20, 2025. The March 20 framework is its replacement. EO 14110, for all its limitations, at least acknowledged displaced workers as a policy subject. The March 20 framework acknowledges them as the sixth of six priorities, in a section organized around training for "an AI-powered economy" rather than protection from structural displacement.

The distinction matters. Training workers for an AI-powered economy means teaching existing workers to use AI tools — the skills-gap diagnosis. Protecting workers from structural displacement would mean addressing the income collapse that precedes any successful retraining, the mismatch between the occupations being eliminated and the occupations being created, and the timing problem: retraining programs take two to four years, and the Tufts Fletcher AI Jobs Risk Index projects that 4.9 million workers in "tipping point" occupations could see displacement risk jump from under 10% to over 40% within that same window.

None of this appears in the framework.

What preemption without replacement means

The legal logic of the framework's preemption provision is straightforward: inconsistent state regulation creates compliance costs for AI developers operating across multiple jurisdictions, which disadvantages U.S. firms in the global AI race. This argument is coherent. It is also an argument that federal worker protections should replace state worker protections, not that no worker protections should exist.

The framework does not make that substitution. It proposes to preempt state-level protections — the AI transparency and algorithmic discrimination laws that represent the only enacted regulatory infrastructure in this area — while offering, at the federal level, an aspiration that Congress will eventually fund skills training programs.

What states have enacted, the federal government is proposing to remove. What the federal government is proposing in its place is a suggestion.

The asymmetry

Every major component of the White House framework on the industry side has an enforcement structure. Intellectual property rights are legally enforceable. Anti-censorship provisions are tied to existing First Amendment jurisprudence. Preemption, if enacted, is a binding legal standard. Data center permitting streamlining has regulatory teeth.

The workforce objective has none. It encourages Congress. It expresses a desire. It uses the word "creating" as if job creation in an AI economy were something the government is proposing to do, rather than something it is expressing hope the market will eventually accomplish on its own.

This asymmetry is not incidental. It reflects a choice about what the framework is for. The document was not written to protect workers from the labor market effects of AI deployment. It was written to accelerate AI deployment by removing regulatory friction. The sixth objective exists to demonstrate awareness of the labor question, which is different from addressing it.

Demonstrating awareness of a problem without addressing it is a familiar form of policy. It produces text. It does not produce outcomes.

*Sources: White House, "President Donald J. Trump Unveils National AI Legislative Framework," March 20, 2026; White House, National Policy Framework for Artificial Intelligence Legislative Recommendations, March 20, 2026; Tufts Fletcher School / Digital Planet, American AI Jobs Risk Index, March 2026; WilmerHale client alert, "White House Releases National Policy Framework for Artificial Intelligence," March 23, 2026.*