The White House released its national AI legislative framework this week, and the headlines wrote themselves: “Trump unveils AI plan,” “US sets out AI priorities,” “Washington responds to China AI threat.” All technically accurate. All somewhat misleading about what actually happened.

What the Framework Actually Is

This is a congressional framework proposal, not enacted legislation. The White House is laying out principles and priorities it wants Congress to legislate on. Whether Congress passes anything is an entirely separate question.

What the framework does propose:

  • Preemption of state AI laws — a federal floor that prevents states from imposing stricter AI rules
  • Promotion of “AI development” as a national priority across government agencies
  • National security carve-outs around AI and defence/infrastructure
  • Light-touch regulatory approach for most commercial AI applications

The goal, as the framework states, is to “win the AI race” and ensure American AI dominance. The framing is explicitly competitive — against China, against the EU’s regulatory approach.

The State Preemption Clause

The most significant concrete proposal is barring states from enacting their own AI laws. California tried this with its AI safety bill (SB 1047) and faced enormous lobbying pressure. The White House framework effectively endorses that lobbying outcome at the federal level.

This is a major shift. US tech policy has historically been characterised by state-level experimentation — California’s privacy laws, for example, became a de facto national standard. If the federal framework preempts state AI laws, that experimentation ends.

Environmental and civil liberties advocates have particular reason to be concerned. States like California and New York have been the primary source of AI accountability legislation in the absence of federal action.

What It Doesn’t Address

The framework is notable for what it largely ignores:

  • Copyright and training data — silent on the legal exposure of AI companies that trained on copyrighted material
  • Worker displacement — no mention of retraining programmes or labour protections
  • AI in hiring, lending, healthcare — no specific guidance on high-stakes AI decisions affecting individuals
  • Open source AI — no clear policy on whether open-weight models get different treatment

The emphasis is on speed and competitiveness. The assumption is that removing friction will produce better AI outcomes. Whether those outcomes benefit ordinary Americans is treated as someone else’s problem.

The Sanders/AOC Counter

Senator Sanders and Representative Ocasio-Cortez introduced a bill this week requiring a moratorium on new AI-focused data centre construction in the US, citing energy consumption and environmental concerns. It’s a more radical response than anything in the White House framework.

Neither bill is likely to become law in its current form. But the fact that both are on the table tells you something about the range of AI politics currently active in Washington.

The Take

The White House framework is best read as a positioning document — a signal to the AI industry that the administration is on their side, a signal to China that the US isn’t backing down on AI, and a signal to Congress that the President wants action.

Whether it produces actual legislation, and what that legislation looks like, will depend on congressional dynamics that the framework doesn’t control. The AI race continues. The rules are still being written.

Sources: The White House, CNN