AI regulation

Trump and Big Tech lose in the latest jockeying over AI regulation

The threat of a freeze on states enforcing AI laws prompted lawmakers to get creative, forming some unusual political alliances.

By Chase DiFeliciantonio

Allies of President Donald Trump have fallen short in an effort to prevent states from regulating artificial intelligence — the tech industry’s first major defeat under his administration and a prelude to future battles around who gets to write the rulebook on a technology that is rapidly remaking the economy.

After months of intense lobbying on both sides, the Senate voted in the pre-dawn hours Tuesday to defeat a provision in the Republican megabill to bar states from regulating AI over the next 10 years.

The main driver of the measure, Texas Republican Sen. Ted Cruz, has threatened to bring the measure back, suggesting the fight is far from over.

“We have the idea of the federal preemption sort of hanging over our heads,” California state Sen. Jerry McNerney, who has authored numerous AI bills in Sacramento and was vocal about regulating the technology during his years in Congress, told POLITICO. “I don’t think that’s going to go away.”

The freeze became a rallying cry for much of the tech industry and their moneyed backers, who have long argued a patchwork of state AI rules will hobble the United States’ ability to compete with rivals like China when it comes to the technology — an argument that has resonated with the Trump administration and its MAGA allies.

Even without the AI freeze, Trump “is fully supportive of the Senate-passed version of the One, Big, Beautiful Bill and looks forward to House Republicans uniting to pass the bill and send it to his desk by July 4,” said White House spokesperson Abigail Jackson in a statement to POLITICO.

States are already gearing up for the next battle, with legislators crafting measures they hope will be immune to Washington’s preemption trump card.

The push for a federal freeze on state AI laws also forged unlikely bipartisan coalitions across red and blue states that helped to crater Cruz’s first effort. Those alliances will still be there should he try again.

Ten years is “a lifetime in technology,” Utah state Rep. Doug Fiefia, a Republican and former Google employee, said last week during a press blitz by other GOP state officials opposed to the measure.

A moratorium “really hurts not only states and states’ rights to act but … consumers.” he said. “That’s a non-starter for us.”

That’s the view of some Democrats who opposed the moratorium too.

“This was genuinely bipartisan from day one,” North Carolina Attorney General Jeff Jackson, a Democrat and one of 40 state attorneys general who signed a letter decrying the moratorium, told POLITICO before the measure was killed.

“If [lawmakers] try to shackle us and keep us from doing anything in the space, that’ll make things much, much more difficult,” Jackson said. “But I just can’t imagine a world in which this debate turns off.”

With the battle won, the war begins

The House passed a version of the megabill that included a 10-year moratorium in May. Cruz then tied the idea of an AI moratorium in the Senate version to billions of dollars in federal broadband funding, and later millions in AI development money, conditioned on states not enforcing their own AI rules. The idea faced pushback from some other Senate Republicans almost immediately. Cruz’s attempts to rewrite it and strike a last-minute deal failed before the language was voted out for good.

(Cruz didn’t respond to questions on Tuesday about whether he would bring the measure back in the future, but praised the passage of the megabill in a press release that day without mentioning the scrapped moratorium.)

Hundreds of state legislators from Arkansas to Pennsylvania to Texas signed onto a bipartisan letter opposing the moratorium shortly before Cruz introduced a revised version in the Senate in early June, galvanized by the possibility of losing years of work after passing bills on everything from restricting robot doctors to protecting dead celebrities’ images and consumer privacy.

State Democrats even found themselves on the same side as frequent sparring partners like Rep. Marjorie Taylor Greene (R-Ga.) and Sen. Josh Hawley (R-Mo.), who swore to oppose the AI measure, arguing the issue should be handled in a separate bill, and raising concerns it would roll back critical consumer protections.

A deep-pocketed tech industry with close ties to Trump quickly arrayed against that coalition, however, trying to seize the opportunity to build the technology as they saw fit, unfettered by state rules. AI chip-making giant Nvidia was among the first to openly support the moratorium.

Venture capitalists across Silicon Valley, including Andreessen Horowitz, which has been influential in the Trump administration’s AI thinking, framed a freeze as the ticket to winning the AI race. And national trade groups like TechNet and NetChoice kept up the drumbeat of support, even as they continued to lobby on state bills.

State lawmakers have been mulling how to retool some of their own bills to deal with the continuing threat of Congress big-footing their efforts.

California’s McNerney said incentivizing, instead of forcing, companies to test their AI programs could avoid conflicts with Congress and tech companies averse to mandatory restrictions, for example.

Jason Elliott, a former policy adviser to California Gov. Gavin Newsom, outlined another approach: studying the impacts of AI on industries like health care, and update rules that govern those areas instead.

“There’s a lot we can be doing in the immediate term on the immediate impacts on job disruption, on protecting consumers, protecting civil liberties,” when it comes to AI, said Elliott, now an independent political consultant, in an interview with POLITICO.

Putting transparency requirements on AI companies is another approach that California state Sen. Scott Wiener is taking with a bill this year. His measure includes expanded whistleblower protections for AI workers, which build on existing legal protections that aren’t specifically about AI.

Other state lawmakers across the country focused on AI legislation are forging ahead without much heed for what Congress is, or is not, doing.

“I’m of the view to say ‘Forget Congress, it’s in our jurisdiction,’” New York state Sen. Andrew Gounardes told POLITICO. The Democrat’s RAISE Act requires large AI developers to disclose when programs go awry and to submit to outside audits to prevent worst-case scenarios like death or property damage.

“We’re going to do this until you force us not to,” he said.

What tech is saying

As much as the issue united some Democrats and Republicans, it forged a unified front among major tech players, with some notable exceptions.

Some marquee firms and their VC backers have lauded the idea of taking away states’ AI regulatory powers as the path to simpler, unified rules governing the technology.

OpenAI’s Global Affairs Chief Chris Lehane argued before the measure was stripped out that, “Across the country, we’re seeing state bills that, while generally well-intentioned, risk creating a maze of conflicting rules without meaningfully improving the safety of frontier AI systems” that could create a barrier to entry for new AI startups.

National tech lobbying groups NetChoice and TechNet had also applauded the moratorium, saying it avoids a state-by-state approach to AI regulation, and gives Congress time to come up with a unified framework.

Tech’s moneyed backers have also spoken up.

Collin McCune, a lobbyist for the venture capital firm Andreessen Horowitz, threw his support behind the moratorium, while Bobby Franklin, president and chief executive of the National Venture Capital Association, wrote a letter to Senate Majority Leader John Thune supporting it.

Some in tech warned against a moratorium with no legal framework to back it up.

The CEO of Anthropic, which developed a competitor to ChatGPT, Claude, wrote in a recent op-ed that while he was sympathetic to the patchwork argument, a sweeping stoppage of AI regulation without a national transparency-focused legislative standard to fill the void could have disastrous consequences.

“Without a clear plan for a federal response, a moratorium would give us the worst of both worlds — no ability for states to act, and no national policy as a backstop,” Dario Amodei wrote in the New York Times.

That doesn’t mean Anthropic is pushing for just any state legislation.

The company’s co-founder Jack Clark voiced concern with Gounardes’ bill in New York without taking a position on it, calling it overly broad. “If there isn’t anything at the federal level, we’ll continue to engage on bills at the state level,” Clark wrote.

In response to tech criticism of what companies see as a patchwork system, state lawmakers note that many of their bills emulate one another, undermining the idea that there are vast differences companies must contend with. Washington state Rep. Clyde Shavers told POLITICO his two AI-focused bills on AI model transparency and notifying users when online content is AI- generated closely resemble California measures.

Strange bedfellows

Lawmakers, attorneys general and Republican governors from both parties had deluged Congress with letters opposing the moratorium, and tacit threats to sue over it.

Similarly, a group of five Republican state lawmakers from Montana, Utah, South Carolina Tennessee and Ohio, including Utah’s Fiefia, recently spoke out against the moratorium, calling it a violation of the U.S. Constitution’s 10th amendment and an unfair usurpation of federalism.

“The way this moratorium is written, it opens up a legal minefield,” said Montana state Sen. Barry Usher during the briefing last week. “It’s vague. It invites lawsuits. It leaves states guessing about what they can and can’t do.”

While the state attorneys general who spoke out against the provision did not explicitly threaten to sue, they argued the measure could harm consumers, an area where they have stepped in before to challenge alleged privacy and social media harms.

That broad, bipartisan pushback has “made a huge difference in this debate,” said Jason Van Beek, a former longtime adviser to Thune, before the measure failed. Van Beek now works as chief government affairs officer with the nonprofit Future of Life Institute, which focuses on potential AI risks.

Republican California Assemblymember Carl DeMaio — a firebrand Trump supporter in the deep blue state — took the unusual move of joining state Democrats in signing one of the bipartisan letters opposing the moratorium.

“My opposition to the moratorium is based more on an absence of federal action,” DeMaio told POLITICO before the measure was stripped out.

“When Congress doesn’t act, then a state needs to look for ways to act.”

He argued Democrats and Republicans coming together to pass good AI laws is the best antidote in future moratorium fights, pointing to a bill he authored this year as a common-sense approach. That bill would require AI businesses to obtain consent from California-based users before sharing or storing their data in foreign countries

Newsom has made similar remarks about existing state laws, underlining the unlikely bipartisanship stemming from the AI moratorium debate. His office pointed to how the freeze could undermine existing state laws “that ban AI-generated child pornography, deepfake porn, and robocall scams against the elderly.”

While state AGs implied they might sue over the moratorium, Jackson, the North Carolina attorney general, had another solution.

“There’s an easy fix here,” he said.

“All Congress has to do is pass a bill putting up some AI safeguards for consumers and voters before they pass a bill that tries to ban all of ours.”