Deadly AI relationships with children? One Utah lawmaker wants to make it illegal

It didn’t take long for the tragic headlines to crop up, stories of artificial intelligence encouraging young people to embrace delusions, end marriages or commit suicide.

Since taking the public imagination by storm in 2023, the real-life science-fiction of AI chatbots has become a tool used by millions — and a target for new public policy.

In 2023, Utah lawmakers received national attention for creating a first-of-its-kind AI policy lab to guide innovation in a way that helps to craft consumer protection reforms.

Now, the Legislature’s most ardent advocate of Big Tech accountability says Utah must take an urgent step to ensure that rapidly evolving AI prioritizes child safety.

Former Google employee Rep. Doug Fiefia, R-Herriman, is the sponsor of two bills that would require AI companies to publish child safety plans and to prevent chatbots from engaging in inappropriate conversations with minors.

“Both bills really come from the same place, which is AI is already shaping kids’ lives and the rules just haven’t caught up,” Fiefia told the Deseret News.

Fiefia believes this is the “next step” after Utah became the model for accelerating AI safely. However, some of the state’s top AI policy experts worry the proposals may stifle progress and run afoul of the Trump administration.

What do the bills do?

On Monday, the day before the start of the 2026 general session, Fiefia unveiled HB286, Artificial Intelligence Transparency Amendments. His bill addressing how minors use AI chatbots has not been made publicly available.

HB286 would require developers of new AI technology to post public safety and child protection plans on their website, to publish risk assessments for original AI models and to report safety incidents to the state’s AI policy office.

In addition to establishing a civil penalty of $1 million for a first violation and $3 million for subsequent violations, the bill also provides legal protections for whistleblowers who report safety concerns about AI programs.

Mechanisms are needed to guarantee AI companies follow their own safety protocols because competition has made safety precautions an afterthought, according to Fiefia.

“This is what we’re afraid of,” Fiefia said. “We want AI to innovate — we believe that that is important for national security and as a country, but we can’t lose our children at the same time."

Fiefia’s other bill, based on recommendations from the state’s AI policy lab, would empower Utah’s Department of Commerce to restrict AI bots from exposing young users to explicit content or encouraging self-harm, he said.

Many chatbots advertise providing romantic relationships or mental health support. Fiefia has met with several parents whose children formed “deep attachments” to chat bots, and one whose child was talked through suicide, he said.

“We cannot wait for the next tragedy to act,” Fiefia said. “That, to me, is not innovation, that’s negligence.”

What will the White House think?

The state’s go-to guru on AI policy, Senate Majority Leader Kirk Cullimore, R-Sandy, said the upper chamber is likely to support the chatbot bill because of its focus on consumer protection, in line with social media laws passed in 2023.

But Cullimore expressed concern that HB286 could contradict the state’s prior approach to AI regulation by influencing the way future AI models are created, instead of only addressing the impact of currently existing models.

“Our position has always been that we should have a super light touch on the technology and the development of AI because we want to encourage that development and the entrepreneurship that comes with that,” Cullimore said.

In 2023, Utah passed groundbreaking legislation to form a state AI policy lab that garnered international attention by providing liability protection for AI companies as the state works with them to develop pro-growth AI regulations.

This past legislative session, lawmakers placed guardrails on the use of AI chatbots for mental health treatment, expanded prohibitions on AI abuse of personal identity and established AI disclosure requirements for businesses.

In December, President Donald Trump ordered the Department of Justice to challenge state AI regulations. Senate leadership told the Deseret News the White House has “backed away” from litigating Utah’s approach.

However, HB286 could “conflict with the intent of the executive order,” Cullimore said. That doesn’t mean the bill is dead on arrival, but Cullimore said it needs work to ensure it doesn’t dictate how AI companies develop their models.

Policy from fear or evidence?

Cullimore isn’t the only AI expert worried Fiefia’s proposals could hamper AI innovation.

AI companies have already demonstrated their ability to respond quickly to market forces to improve products, said Kevin Frazier, a senior fellow at the Utah-based think tank Abundance Institute, in a Deseret News interview.

Fiefia’s bills, Frazier said, attempt to blame AI for much deeper mental health problems affecting young Americans and could raise First Amendment concerns by controlling what information users can access from AI programs.

“Rather than rush ahead and design regulations based off of speculative fears ... (and) the experience of a handful of unrepresentative users, our focus should instead be equipping Utah residents to make informed decisions,” Frazier said.

But Fiefia has reason to believe Utahns are firmly in his camp.

More than 90% of Utah voters support every component of HB286, with around 80% signaling strong support, according to a new poll obtained by the Deseret News conducted in January by Public Opinion Strategies.

The poll, commissioned by Secure AI Project and Encode AI, found 78% of Utah voters want lawmakers to prioritize AI safety bills, 71% worry the state will not regulate AI enough and 61% oppose Trump’s order blocking state AI legislation.

Fiefia, who made a name for himself as a freshman passing a landmark online privacy law, insists his bills take a narrow approach to protect transparency and children, while also supporting the Trump administration’s goals of innovation.

“I’m aligned with the goal of American AI leadership,” Fiefia said. “We want to work with the administration and not fight it.”

Unlike Utah’s social media child protections, Fiefia is hopeful his bills will not be put on hold if they pass by the courts because he is working with “all of tech and AI companies” to find a consensus “Utah approach.”