WASHINGTON — The Trump administration on Thursday unveiled an artificial intelligence policy framework designed to do something no federal agency has managed yet: establish a single national standard for AI governance that would effectively neutralize the patchwork of state-level regulations spreading across the country. The move represents the most aggressive federal assertion of authority over artificial intelligence to date, and it arrives at a moment when state legislatures from Sacramento to Albany are racing to fill what they see as a regulatory vacuum.
The policy, announced by the White House Office of Science and Technology Policy, doesn’t carry the force of law on its own. Not yet. But it lays the groundwork for executive action and potential legislation that would preempt state AI regulations, arguing that a fragmented regulatory environment threatens American competitiveness and innovation. The framework calls for uniform federal standards governing AI safety, transparency, and deployment — standards that would supersede the more than 400 AI-related bills introduced in state legislatures over the past two years, as reported by The New York Times.
“We cannot allow fifty different AI regimes to strangle the most important technology of our lifetime,” said Michael Kratsios, the administration’s chief technology officer, during a briefing at the Eisenhower Executive Office Building. “American AI leadership depends on regulatory clarity, not regulatory chaos.”
The timing is no accident.
California Governor Gavin Newsom vetoed a sweeping AI safety bill last year, but the state legislature has already introduced successor legislation. Colorado enacted its own AI discrimination law. Illinois has moved aggressively on AI in hiring. Texas, Connecticut, and New York have all advanced proposals targeting everything from deepfakes to algorithmic decision-making in housing and insurance. The result is exactly the kind of jurisdictional mess that technology companies have long warned about — and that the White House is now using as justification for federal intervention.
Silicon Valley’s largest companies have responded with barely contained enthusiasm. The Information Technology Industry Council, whose members include Google, Microsoft, Amazon, and Meta, released a statement within hours calling the framework “a welcome and overdue step toward coherent national AI policy.” The lobbying group BSA | The Software Alliance echoed the sentiment, arguing that state-by-state regulation imposes compliance costs that disproportionately burden smaller AI firms while doing little to address genuine safety concerns.
But the applause from industry hasn’t been universal, and the resistance from state officials has been swift and sharp. California Attorney General Rob Bonta called the framework “a transparent attempt to shield Big Tech from accountability” in a statement posted to social media. New York State Senator Andrew Gounardes, who has sponsored AI transparency legislation, told reporters that the White House plan amounts to “deregulation dressed up as regulation.”
The tension between federal preemption and state authority is, of course, nothing new. Telecommunications. Financial services. Data privacy. The same fight has played out across industries for decades. What makes AI different is the speed at which the technology is being deployed and the breadth of sectors it touches. An AI system that screens job applicants in one state, approves mortgage applications in another, and generates medical diagnoses in a third could theoretically be subject to dozens of overlapping and potentially contradictory legal requirements. The compliance burden is real. So is the risk that weaker federal standards could effectively deregulate an industry that many consumer advocates believe needs more oversight, not less.
The White House framework, as described by The New York Times, includes several key provisions. It would establish a federal AI risk classification system, loosely modeled on the European Union’s AI Act but with significantly lighter compliance requirements for most applications. It would mandate that companies deploying “high-risk” AI systems — those used in criminal justice, healthcare, employment, and financial services — conduct impact assessments and maintain documentation of training data and model performance. And it would create a federal preemption clause, explicitly barring states from imposing requirements on AI systems that exceed federal standards.
That last provision is the one drawing the most fire.
“Preemption is a loaded gun,” said Alondra Nelson, the former deputy director of the White House Office of Science and Technology Policy under President Biden. “Once you tell states they can’t act, you’d better be confident that the federal government will. And this administration’s track record on technology enforcement doesn’t inspire that confidence.”
Nelson’s skepticism points to a fundamental asymmetry in the debate. States have been acting precisely because the federal government hasn’t. Congress has failed to pass comprehensive AI legislation despite years of hearings, proposed bills, and bipartisan expressions of concern. The AI Research, Innovation, and Accountability Act, introduced with fanfare in 2023, never made it to a floor vote. Neither did the Algorithmic Accountability Act. The SAFE Innovation Framework championed by Senate Majority Leader Chuck Schumer produced a series of forums and a set of principles but no binding legislation.
Into that void, states moved. And now the White House is telling them to stop.
The administration argues that its framework isn’t deregulatory — that it establishes meaningful baseline standards while preventing the fragmentation that would ultimately harm consumers and innovators alike. Kratsios pointed to the example of autonomous vehicles, where a lack of federal standards led to a confusing patchwork of state rules that he said slowed deployment and created safety inconsistencies. “We learned that lesson the hard way,” he said. “We’re not going to repeat it with AI.”
Industry groups have been laying the groundwork for this moment for months. OpenAI, which has faced regulatory pressure in multiple states and countries, published a policy paper in January arguing for federal preemption of state AI laws. The company’s head of global affairs, Chris Lehane, has been making the rounds in Washington, meeting with congressional staff and administration officials. Anthropic, OpenAI’s chief rival, has taken a somewhat more nuanced position, calling for strong federal standards that would make state action unnecessary rather than explicitly illegal. Google’s Kent Walker wrote in a blog post last month that “innovation requires consistency” and urged Congress to act before the regulatory map becomes unnavigable.
Consumer advocacy groups see the industry’s enthusiasm as confirmation of their worst fears. “When every major tech company is applauding a regulatory framework, that should tell you something,” said Lisa Hayes, president of the Center for Democracy and Technology. “The question isn’t whether we need federal AI standards. Of course we do. The question is whether this framework is designed to protect people or to protect market position.”
The framework does include provisions that consumer groups have long sought — particularly the impact assessment requirements for high-risk systems and a mandate for algorithmic auditing in certain contexts. But critics note that the enforcement mechanisms are vague. The document assigns primary oversight responsibility to existing sector-specific regulators: the FDA for AI in medical devices, the FTC for consumer-facing AI applications, the EEOC for AI in employment. It does not create a new federal AI agency, despite calls from some lawmakers and academics for exactly that.
Senator Richard Blumenthal, who co-authored one of the failed congressional AI bills, called the sector-specific approach “inadequate to the scale of the challenge.” He noted that many AI applications don’t fit neatly into existing regulatory categories. “An AI system that generates synthetic content, influences elections, and displaces workers simultaneously — which agency is responsible for that?” he asked during a press conference on Capitol Hill.
The preemption question also raises thorny constitutional issues. Federal preemption of state law typically requires an act of Congress, not just an executive branch policy statement. The White House framework acknowledges this, calling on Congress to codify its provisions into law. But given the current gridlock on Capitol Hill, that prospect seems remote. In the absence of legislation, the framework’s preemption clause would likely face immediate legal challenges if the administration attempted to enforce it through executive action or agency rulemaking.
“This is a statement of intent, not a statement of law,” said Paul Weiss partner and former DOJ official Andrew Goldsmith. “The real question is what happens next — whether Congress acts, whether agencies issue rules, and whether courts uphold them.”
Meanwhile, state legislators aren’t waiting. The California legislature is advancing a new version of its AI safety bill, this time with narrower scope designed to survive a gubernatorial veto. Colorado’s AI discrimination law takes effect in 2026. Illinois is considering expanding its AI video interview law to cover a broader range of automated employment decisions. And a coalition of state attorneys general from 15 states sent a letter to Congress last week opposing any federal preemption of state AI consumer protection laws.
The letter, obtained by Politico, argued that “states have historically served as laboratories of democracy for technology regulation” and that “premature federal preemption would eliminate the ability of states to protect their residents from AI-related harms that the federal government has not yet addressed.”
That argument carries historical weight. State laws on data breach notification, which began with California’s SB 1386 in 2002, eventually pushed Congress toward considering federal privacy legislation. State consumer protection lawsuits against social media companies have forced disclosures and policy changes that federal regulators didn’t pursue. Proponents of state AI regulation argue that eliminating state authority now would remove the very pressure that might eventually produce stronger federal standards.
The counterargument, advanced by the administration and its industry allies, is that AI is fundamentally different from previous technologies in its speed of development and breadth of application. “You can’t regulate AI the way you regulated telephone companies,” Kratsios said. “The technology changes too fast, crosses too many borders, and defies traditional jurisdictional categories. Federal leadership isn’t optional. It’s essential.”
Both sides are right about parts of this, which is what makes the debate so intractable.
The international dimension adds another layer of complexity. The European Union’s AI Act, which began taking effect in stages last year, has created a comprehensive regulatory framework that applies to any AI system operating in or affecting EU markets. China has implemented its own AI regulations, focused heavily on content generation and algorithmic recommendation systems. The administration has argued that the United States needs a unified domestic framework to negotiate effectively on international AI standards — that America can’t lead globally if it’s fractured internally.
There’s something to that argument. Companies operating across borders need clear rules. And the prospect of American AI firms facing stricter regulation in California than in the EU is not, on its face, a rational outcome. But the solution matters as much as the problem. A federal framework that establishes genuinely protective standards while streamlining compliance is one thing. A framework that uses the language of preemption to lower the regulatory floor is another.
The coming months will determine which version materializes. Congressional leaders from both parties have expressed interest in AI legislation, and the White House framework could serve as a starting point for negotiations. But the details — enforcement mechanisms, private rights of action, the scope of the preemption clause, the definition of “high-risk” — will be fiercely contested. Lobbying expenditures on AI policy have tripled since 2023, according to OpenSecrets data, and every major technology company has expanded its Washington presence.
For now, the framework is a signal. A loud one. The White House has drawn a line: AI regulation belongs in Washington, not in state capitols. Whether that line holds — legally, politically, practically — is the question that will define American AI governance for years to come.
And the states aren’t backing down.
WebProNews is an iEntry Publication