By MATT BROWN and MATT O’BRIEN

WASHINGTON (AP) — House Republicans surprised tech industry watchers and outraged state governments when they added a clause to Republicans’ signature “ big, beautiful ” tax bill that would ban states and localities from regulating artificial intelligence for a decade.

The brief but consequential provision, tucked into the House Energy and Commerce Committee’s sweeping markup, would be a major boon to the AI industry, which has lobbied for uniform and light touch regulation as tech firms develop a technology they promise will transform society.

However, while the clause would be far-reaching if enacted, it faces long odds in the U.S. Senate, where procedural rules may doom its inclusion in the GOP legislation.

“I don’t know whether it will pass the Byrd Rule,” said Sen. John Cornyn, R-Texas, referring to a provision that requires that all parts of a budget reconciliation bill, like the GOP plan, focus mainly on the budgetary matters rather than general policy aims.

“That sounds to me like a policy change. I’m not going to speculate what the parliamentarian is going to do but I think it is unlikely to make it,” Cornyn said.

Senators in both parties have expressed an interest in artificial intelligence and believe that Congress should take the lead in regulating the technology. But while lawmakers have introduced scores of bills, including some bipartisan efforts, that would impact artificial intelligence, few have seen any meaningful advancement in the deeply divided Congress.

An exception is a bipartisan bill expected to be signed into law by President Donald Trump next week that would enact stricter penalties on the distribution of intimate “revenge porn” images, both real and AI-generated, without a person’s consent.

“AI doesn’t understand state borders, so it is extraordinarily important for the federal government to be the one that sets interstate commerce. It’s in our Constitution. You can’t have a patchwork of 50 states,” said Sen. Bernie Moreno, an Ohio Republican. But Moreno said he was unsure if the House’s proposed ban could make it through Senate procedure.

The AI provision in the bill states that “no state or political subdivision may enforce any law or regulation regulating artificial intelligence models, artificial intelligence systems, or automated decision systems.” The language could bar regulations on systems ranging from popular commercial models like ChatGPT to those that help make decisions about who gets hired or finds housing.

State regulations on AI’s usage in business, research, public utilities, educational settings and government would be banned.

The congressional pushback against state-led AI regulation is part of a broader move led by the Trump administration to do away with policies and business approaches that have sought to limit AI’s harms and pervasive bias.

Half of all U.S. states so far have enacted legislation regulating AI deepfakes in political campaigns, according to a tracker from the watchdog organization Public Citizen.

Most of those laws were passed within the last year, as incidents in democratic elections around the globe in 2024 highlighted the threat of lifelike AI audio clips, videos and images to deceive voters.

California state Sen. Scott Wiener called the Republican proposal “truly gross” in a social media post. Wiener, a San Francisco Democrat, authored landmark legislation last year that would have created first-in-the-nation safety measures for advanced artificial intelligence models. The bill was vetoed by California Gov. Gavin Newsom, a fellow San Francisco Democrat.

“Congress is incapable of meaningful AI regulation to protect the public. It is, however, quite capable of failing to act while also banning states from acting,” Wiener wrote.

A bipartisan group of dozens of state attorneys general also sent a letter to Congress on Friday opposing the bill.

“AI brings real promise, but also real danger, and South Carolina has been doing the hard work to protect our citizens,” said South Carolina Attorney General Alan Wilson, a Republican, in a statement. “Now, instead of stepping up with real solutions, Congress wants to tie our hands and push a one-size-fits-all mandate from Washington without a clear direction. That’s not leadership, that’s federal overreach.”

As the debate unfolds, AI industry leaders are pressing ahead on research while competing with rivals to develop the best — and most widely used —AI systems. They have pushed federal lawmakers for uniform and unintrusive rules on the technology, saying they need to move quickly on the latest models to compete with Chinese firms.

Sam Altman, the CEO of ChatGPT maker OpenAI, testified in a Senate hearing last week that a “patchwork” of AI regulations “would be quite burdensome and significantly impair our ability to do what we need to do.”

“One federal framework, that is light touch, that we can understand and that lets us move with the speed that this moment calls for seems important and fine,” Altman told Sen. Cynthia Lummis, a Wyoming Republican.

And Sen. Ted Cruz floated the idea of a 10-year “learning period” for AI at the same hearing, which included three other tech company executives.

“Would you support a 10-year learning period on states issuing comprehensive AI regulation, or some form of federal preemption to create an even playing field for AI developers and employers?” asked the Texas Republican.

Altman responded that he was “not sure what a 10-year learning period means, but I think having one federal approach focused on light touch and an even playing field sounds great to me.”

Microsoft’s president, Brad Smith, also offered measured support for “giving the country time” in the way that limited U.S. regulation enabled early internet commerce to flourish.

“There’s a lot of details that need to be hammered out, but giving the federal government the ability to lead, especially in the areas around product safety and pre-release reviews and the like, would help this industry grow,” Smith said.

It was a change, at least in tone, for some of the executives. Altman had testified to Congress two years ago on the need for AI regulation, and Smith, five years ago, praised Microsoft’s home state of Washington for its “significant breakthrough” in passing first-in-the-nation guardrails on the use of facial recognition, a form of AI.

Ten GOP senators said they were sympathetic to the idea of creating a national framework for AI. But whether the majority can work with Democrats to find a filibuster-proof solution is unclear.

“I am not opposed to the concept. In fact, interstate commerce would suggest that it is the responsibility of Congress to regulate these types of activities and not the states,” said Sen. Mike Rounds, a South Dakota Republican.

“If we’re going to do it state by state we’re going to have a real mess on our hands,” Rounds said.

O’Brien reported from Providence, Rhode Island. AP writers Ali Swenson in New York, Jesse Bedayn in Denver, Jeffrey Collins in Columbia, South Carolina, and Trân Nguyễn in Sacramento, California contributed to this report.