State parliamentarians examine AI’s promises and pitfalls ahead of November elections

LOUISVILLE, Kentucky – In a white-walled conference room, a speaker surveyed hundreds of state lawmakers and policymakers, asking them whether artificial intelligence posed a threat to elections in their states.

The results were clear: 80 percent of participants in a live poll said “yes.” In a follow-up question, nearly 90 percent said their state’s laws were not sufficient to ward off these threats.

The topic was among many discussions on artificial intelligence that dominated meetings last week of the National Conference of State Legislatures, the largest annual gathering of lawmakers, in Louisville.

“This is the topic of the day,” Kentucky Republican Senator Whitney Westerfield told lawmakers as he opened one of many panels on AI. “There are a lot of discussions going on about this in all of our legislatures across the country.”

While some experts and politicians praised the benefits of AI for healthcare and education, others lamented that it could disrupt the democratic process just months before the November election. And politicians compared the many legislative proposals they are proposing to address the problem.

This presidential election cycle is the first since generative AI — a form of artificial intelligence that can create modern images, audio and video — became widely available. This has raised alarm about deepfakes, remarkably convincing but counterfeit videos or images that can portray any person, including candidates, in situations that did not happen or saying things they did not say.

“We need to do something to make sure voters understand what they are doing,” said Kentucky State Senator Amanda Mays Bledsoe.

The Republican congressman, who chairs a special legislative working group on AI, was co-initiator of a cross-party bill this year aimed to restrict the exploit of deepfakes to influence elections. The bill would have allowed candidates whose appearance, behavior, or language in an election communication was altered by “synthetic media” to sue the sponsor. The state Senate unanimously approved the proposal, but it failed to advance in the House of Representatives.

While Bledsoe expects to bring up the bill again next session, he acknowledged the complexity of the issue: Lawmakers are trying to balance the risks of evolving technology with their desire to encourage innovation and protect free speech.

“You don’t want to go too fast,” she said in an interview, “but you don’t want to be too far behind either.”

Democratic Rhode Island State Senator Dawn Euer told Stateline she is concerned about AI’s potential to amplify disinformation, especially on social media.

“Election propaganda and disinformation have been part of the zeitgeist since the dawn of humanity,” said Euer, chairman of the Senate Judiciary Committee. “Now we have high-tech tools to do it.”

Connecticut State Senator James Maroney, a Democrat, agreed that concerns about AI’s impact on elections are valid, but stressed that most deepfakes target women with digitally generated, non-consensual intimate images or revenge porn. Research firm Sensity AI has been tracking online deepfake videos for years. 90% of it is non-consensual pornwhich are mainly aimed at women.

Maroney sponsored a bill this year that would have regulated artificial intelligence and criminalized deepfake porn and false political messages. This bill passed the state Senate, but not the House of Representatives. Democratic Governor Ned Lamont rejected the measureand said it was premature and potentially damaging to the state’s technology industry.

While Maroney has concerns about AI, he says the benefits far outweigh the risks. For example, AI can lend a hand lawmakers communicate with their constituents through chatbots or translate messages into other languages.

Top election officials on AI

During a meeting in Louisville, New Hampshire Republican Secretary of State David Scanlan said AI could improve election administration by making it easier to organize election statistics or communicate official messages to the public.

Nevertheless, New Hampshire experienced some of the disadvantages of the modern technology first hand earlier this year when Voters received robocalls The program used artificial intelligence to imitate the voice of President Joe Biden to discourage participation in the January primaries.

Public prosecutor charged the political activist who allegedly linked the counterfeit calls to more than a dozen crimes, including voter suppression, and the Federal Communications Commission has proposed a $6 million fine against him.

Even though the technology is modern, Scanlan said election officials have always had to keep a watchful eye on misinformation about the election and extreme tactics by candidates or their supporters and opponents.

“You could call them dirty tricks, but they have always been part of the candidates’ repertoire, and this was indeed a form of that,” he said. “It’s just more complex.”

The response of state officials – they quickly identified the calls as counterfeit and investigated their origin – serves as a model for other states ahead of the November election, said Cait Conley, a senior adviser at the U.S. Cybersecurity and Infrastructure Security Agency who focuses on election security.

“What we’ve seen in New Hampshire is best practice,” she said during the presentation. “They’ve been quick and clear and provided guidance and really just fact-checked the disinformation that’s circulating.”

Kentucky Republican Secretary of State Michael Adams told Stateline that AI could pose a challenge for swing states in the presidential election, but he said the technology may still be too modern to cause widespread problems in most states.

“Of the 99 things we bite our nails about, it’s not in the top 10 or 20,” he said in an interview. “I don’t know if it’s mature enough to be used everywhere.”

Adams received this year’s John F. Kennedy Profile in Courage Award for the commitment to the integrity of the elections despite resistance from his Republican colleagues. He said AI is another obstacle for election officials who already face challenges like disinformation and foreign influence.

Further invoices to follow

Since Congress did nothing, States have increasingly tried to regulate tackle the rapidly evolving world of AI on their own.

NCSL followed this year AI bills in at least 40 states, Puerto Rico, the Virgin Islands and Washington, DC

As states examine the problem, many are looking to Colorado, which this year became the first state to create a comprehensive regulatory framework for artificial intelligence. Technology companies opposed the measure, fearing it would hamper innovation in a modern industry.

Colorado Senate Majority Leader Robert Rodriguez, a Democrat who sponsored the bill, said lawmakers largely took their cues from European Union rules in their wording to avoid improper rules for companies that exploit AI. Still, the bill will be reviewed by a legislative working group before it takes effect in 2026.

“It’s a unique bill in the country and I have no illusions that it’s perfect and ready to go,” he said. “We have two years.”

When Texas lawmakers reconvene next January, state Rep. Giovanni Capriglione expects many artificial intelligence bills to be on the table.

Capriglione, a Republican and co-chair of a state advisory panel on artificial intelligence, expressed concern about the impact of generative artificial intelligence on how people vote – or whether they vote at all – in local and federal elections.

“Artificial intelligence is undoubtedly being used to spread disinformation and misinformation,” he said. “And I think the closer we get to the election, the more cases of this use we will see.”