Two state lawmakers weigh how to curb deceptive AI content without silencing political speech.
State Legislature News Staff, National Conference of State Legislatures
A fake video can spread faster than a correction.
Thirty-one states have enacted laws regulating AI and deepfakes in campaigns and elections. The challenge: stopping fabricated ads without restricting political speech.
In the latest installment of “State of Play,” a bipartisan video series created by A Starting Point and NCSL, California Assemblymember Gail Pellerin (D) and Arizona Rep. Justin Wilmeth (R) discuss what their states require and what happens when a fake reaches voters before anyone can stop it.
Pellerin draws a line between improving a video’s lighting and creating a realistic depiction of a candidate accepting a bribe.
“AI itself is not the problem,” she says. “The problem is when these tools are used to create something materially false that looks real.”
Wilmeth says misleading advertising predates AI, but the technology makes responding harder. A campaign can release a deceptive video, remove it the next day and apologize after the harm has occurred.
“Of course, we’ve got to come up with the right regulations,” he says, “but all it takes is one campaign to put out one video.”
When a Fake Goes Public
Arizona requires disclosures on AI-manipulated content posted within 90 days of an election, Wilmeth says. Candidates also can ask a judge to declare a manipulated photo or video fake, though that process does not allow removal orders or fines solely for creating it.
California’s approach covers deceptive content, platform responsibility and AI disclosures, Pellerin says. The state does not ban AI in campaigns, she adds, but its laws target realistic fakes that deceive voters.
Both point to social media as an enforcement challenge. Pellerin says online platforms do not always accommodate disclosures the way traditional advertisements do. Wilmeth questions how restrictions would apply to an unpaid individual creating political content independently of a campaign.
“Political speech receives strong constitutional protection,” Pellerin says. “So these laws have to be carefully and narrowly written.”
Wilmeth says candidates can use existing libel and defamation laws to challenge false AI videos in court. He says he favors penalties strong enough to discourage violations without becoming debilitating.
“I don’t want to live in a country where even somebody that might denigrate me doesn’t have the ability to say what they want,” he says.
Pellerin emphasizes rapid intervention rather than punishment. For materially deceptive election content, she says, an important remedy is asking a judge to stop its distribution.
“In California, our goal has been compliance and deterrence, not necessarily punishment,” she says.
