White House Pushes National AI Rules

The White House with fountain and red flower bed — Photo: Luca Perra / Shutterstock

The White House is moving to block tough state AI rules while inviting companies to share powerful models for voluntary federal reviews, tightening Washington’s grip without adding firm guardrails.

Story Highlights

President Trump set a “minimally burdensome” national AI policy and moved to preempt strict state laws.
A June 2026 order asks companies to voluntarily share frontier models for classified cybersecurity checks.
Critics say the approach lacks clear enforcement and leaves the public exposed to emerging AI risks.
States have surged ahead on AI bills, setting up a federal–state showdown over who sets the rules.

What the Administration Changed on AI

President Trump signed orders in January and December 2025 to clear federal obstacles and set a national framework that limits new burdens on artificial intelligence. The January action directed agencies to revise or rescind rules seen as blocking U.S. leadership. The December order promoted a single national approach and prepared the Department of Justice to challenge state rules that could hinder innovation. The White House tied this stance to jobs, growth, and national security competition.

In July 2025, the White House pushed procurement toward “unbiased” tools and unveiled America’s AI Action Plan, presented as a roadmap to win the global race. The Office of Management and Budget followed with guidance to speed federal adoption of artificial intelligence across agencies. These moves show a focus on using artificial intelligence in government and the economy, not building a heavy new rulebook. Supporters argue light-touch rules keep the United States ahead of China.

What the New 2026 Order Adds—and What It Does Not

In June 2026, a new order emphasized innovation and security and asked companies to grant voluntary pre-release access to certain frontier models for up to 30 days so the government can run cybersecurity tests. The order also created an artificial intelligence security clearinghouse to help critical infrastructure operators find and fix software risks. These are coordination tools. They do not create mandatory licensing or a liability scheme if systems fail or cause harm, which some analysts have flagged.

Representative Don Beyer, a leader on the Congressional artificial intelligence caucus, called the approach underwhelming. He said it lacks a credible framework to manage risks and still lets companies release powerful models without meaningful review. Civil liberties advocates argued the strategy pares back safety steps in the name of speed and leaves gaps on threats like cyberattacks on government systems. These critics want clear enforcement powers, independent oversight, and defined remedies when people are harmed.

Why States Are Pushing Back

State lawmakers have raced ahead on artificial intelligence. During 2025, all 50 states and several territories introduced bills, and many states enacted measures focused on high-risk uses and discrimination concerns. This state surge set up a clash with the White House plan. The December 2025 framework positioned the Department of Justice to challenge state laws viewed as obstructive and even raised the prospect of tying some federal funds to compliance with a national approach. That step aims to reduce a patchwork, but it also fuels fears of overreach.

For many Americans, the fight echoes a bigger pattern: Washington promises growth while daily risks and costs fall on families and small firms. Skeptics on the right worry about big tech and big government picking winners. Skeptics on the left worry about weak rules for powerful tools used in hiring, health, finance, and policing. Both sides see elites steering policy while regular people shoulder the damage when systems go wrong. The preemption push puts that tension front and center.

What This Means for Consumers, Workers, and Small Firms

Businesses want one clear rulebook so they can build and hire. States want guardrails close to the ground where harms show up first. The current plan favors speed and a single national path. It adds coordination and testing, but it stops short of strict mandates. If a major artificial intelligence failure hits consumers or critical services, pressure will rise fast for tougher rules. If innovation brings clear gains without big harms, the light-touch case gets stronger.

Until Congress passes a durable law, executive orders and agency memos will set the pace. That keeps policy change fast, but also fragile. Companies may face shifting rules across elections, while states keep writing their own. Readers should watch three markers: whether model testing moves from voluntary to required, whether Congress sets clear liability for artificial intelligence harms, and whether courts back the Department of Justice when it challenges state laws.