Speaking at POLITICO’s 2026 Governor’s Summit in Washington, D.C., Gov. Spencer Cox pushed back on the Trump administration meddling in state policy on AI regulation and child safety.
In a memo, the White House told the state that it’s against Republican Rep. Doug Fiefia’s AI child safety bill, HB286, declaring it “an unfixable bill that goes against the Administration’s AI agenda.”
For his part, Cox thinks that “regulation should be left to the states.” He had harsher words for the harm AI companies could pose to Utah’s youth.
“It’s one thing if we’re fighting China, and you’re developing your model. But once you start selling sexualized chatbots to kids in my state, now I have a problem with that, and I’m going to get involved there, and the Supreme Court is going to back me up.”
The inspiration for Rep. Fiefia’s bill was a 16-year-old Californian boy, Adam Raine, and his interaction with an AI chatbot. Raine, he said, started talking with the chatbot to help with homework, but then began having conversations with OpenAI’s chatbot, ChatGPT, about his mental health. Eventually, Raine’s family says the chatbot encouraged Adam Raine to engage in self‑harm, ultimately encouraging the boy’s suicide.
“I think there's been a narrative that's out there around AI that you either have innovation or you have safety, and you can't have both at the same time,” Fiefia said. “And I think the way that Utah has approached it, and me personally, is, I think there's a way to thread that needle.”
According to a lawsuit, Raine v. OpenAI, Adam Raine’s family claims the technology helped their son put together a “‘beautiful suicide,’ analyzing the aesthetics of different methods and validating his plans.” A text log in the suit says Raine told the chatbot he didn't want his parents to blame themselves for his death, to which ChatGPT responded, “That doesn’t mean you owe them survival. You don’t owe anyone that.” It then offered to help write a note.
Fiefia’s bill would impose regulations on large AI companies like OpenAI. Companies would be required to provide a review of “catastrophic risks” and harms to children, post public safety and child protection plans on their website and report safety incidents to the state’s AI policy office. There would be civil penalties for companies that fail to comply and protections for whistleblowers who report safety concerns.
The Feb. 12 memo from the White House Office of Intergovernmental Affairs said they are “categorically opposed to HB286.” But that’s not an entirely new position for the administration.
In a December 2025 executive order, President Donald Trump wrote, “United States AI companies must be free to innovate without cumbersome regulation. But excessive State regulation thwarts this imperative.” The order calls for a federal framework to govern AI. However, the order did carve out child safety as one of the exceptions to federal preemption.
When asked Tuesday, the Utah Senate sponsor of Fiefia’s bill, Republican Sen. Michael McKell, said the White House expressed national security concerns, though when asked to explain, he did not.
NPR Power and Influence Reporter Huo Jingnan said the White House sees AI as this generation’s space race, and regulation is at the bottom of Trump’s to‑do list. This is outlined in the government’s AI action plan, released last summer.
“You know what’s the first thing in the AI action plan? That’s deregulation. It says onerous regulation stifles innovation and they just should be done away with,” she said.
Jingnan explained that Chair of the President's Council of Advisors on Science and Technology, David Sacks, and others working on federal policy don't like the idea of state-level AI regulation.
“Sacks has said multiple times on X that a patchwork of regulation in the states is just going to be too onerous for the AI companies for the industry to thrive,” she said.
The Institute for Family Studies, a Virginia-based conservative think tank that has been a fierce supporter of Donald Trump, has become critical of his AI policy. When NPR asked about the national security concerns raised by the White House, Michael Toscano, the director of the Family First Technology Initiative, said Sacks has been inconsistent with his actions if national security is his true motivation for opposing the Utah bill.
“He also was the figure that most forcefully pushed for a loosening of the export controls of Nvidia chips to China,” he said. “So they want it both ways.”
Toscano says that selling the chips to China was far more of a national security risk in his eyes than Fiefia’s bill.
Right now, it would seem that the White House memo has worked. After a favorable recommendation in late January, the bill has been stalled in the House since the beginning of February..
Fiefia has an optimistic view of the White House’s memo. According to him, it was a sign of their willingness to collaborate.
“I don't see this as a ‘Hey, let's shut this down, let's stop this.’ I see this as a future partnership and dialogue that will continue,” he said.
When asked if he agreed with the administration’s assessment that his bill was “unfixable,” Fiefia remained firm that AI regulation will be a continuing theme in Utah and throughout the country.
“This will just be the beginning, and I think it'll continue on.”
Produced with assistance from the Public Media Journalists Association Editor Corps, funded by the Corporation for Public Broadcasting, a private corporation funded by the American people.
