During Utah’s 2024 governor’s race, a video on social media showed Gov. Spencer Cox admitting to forging signatures. Elsewhere in the same election cycle, voters in New Hampshire received robocalls in the voice of President Joe Biden telling them not to vote in the primaries. And then-billionaire Elon Musk posted a video of former vice president and then-presidential candidate Kamala Harris calling herself a “diversity hire.”
All of these were intended to influence elections. And all of them were generated using artificial intelligence.
Utah lawmakers like Republican Sen. Heidi Balderree want to put a stop to it.
“What I'm after is the, you know, the creation of images that are deceitful with intent to make people believe that a candidate did or said something that they didn't say or do,” she said.
It's too late for Utah to have new rules around AI for this year’s midterms, but Balderree has her eyes on getting a framework in place for 2027 and beyond.
“It's a whole new arena and new territory, so even more so, it's important to get it right,” she said.
Utah already has guidelines when it comes to using AI in campaigns. Currently, you have to disclose if the technology was used when creating content for a political campaign.
That isn’t good enough for Balderree, and she’s looking toward other states like Texas for inspiration. It’s gone a step further to enforce a total ban on candidates using AI-generated content within 30 days of an election.
Balderree likes this idea. She addressed the Government Operations Interim Committee on Aug. 19 about a proposal that includes a ban on realistic AI-generated material intended to injure a candidate 45 days out of an election.
How she crafts the bill will be crucial to whether it gets support. The Trump administration has demonstrated its pro-AI stance and general rejection of regulations on the technology. And if recent history is anything to go by, passing this sort of legislation in Utah could be an uphill battle.
No one knows this better than Republican state Rep. Doug Fiefia.
His attempt to introduce an AI regulation bill this year was quickly snuffed out by a memo from the White House and tabled.
“I think what my takeaway was is that definitions matter,” he said, explaining that when talking about AI, specificity of what is getting regulated is crucial to writing successful legislation.
One of the key concerns around regulating AI in elections is the potential encroachment on the First Amendment, and in particular free speech. Fiefia said balancing civil rights and stopping the spread of misinformation in elections will be difficult.
“It's definitely hard,” he said. “We do want to make sure that what voters see in advertisements and mailers are accurate, and that people aren't using AI to deceive people.”
A 2026 study conducted by the Utah Valley University Center for National Security Studies showed that more than 80% of people can’t tell the difference between a real video and one made by AI, no matter their age or political affiliation.
According to Brandon Amacher, the study lead and program director of the Emerging Tech Policy Program, this is why it’s so important to strike a balance between free speech and intentional misinformation.
AI content is “just as effective at swaying people's opinions as real media is,” he said. His study also demonstrated that a person’s prior AI literacy didn’t help when it came to identifying the difference between real and fake content.
From Amacher’s perspective, one of the more challenging aspects legislatures will face is the speed of AI’s advancement. Policy can become outdated quickly, and lawmakers will forever be playing catch-up.
“I think when you're writing legislation, you need to approach it from the perspective of, hey, we can't fight the tide on this,” he said.
An outright ban would be misguided, said Amacher. Instead, he suggested framing a more open-minded policy that understands the technology’s inevitabilities, like how detailed and accurate AI content is becoming. At the same time, he said it’s crucial to outline the “absolutely unacceptable outcomes,” like content that misleads and lies to the public in crucial spaces like elections.
For Rep. Fiefia, regulating AI in the state is still top of mind heading into the 2027 legislative session, and he said his colleagues are on board, too.
“I think it's going to be a big topic. It continues to move,” he said. “It's obviously a hot topic. I'm actually going to a meeting right now with one of the big AI companies, and so it's going to be something that's widely discussed not only in Utah but across the country.”
