Newsmax is not where you would expect to see an ad for Democratic Senate candidate James Talarico, but there he appeared to be in September, explaining in his own voice, “In the beginning, God was nonbinary and he wasn’t a Christian.”
The video of the former divinity student sitting at a desk and writing in his Bible with a crayon wasn’t real, but a viewer could only be sure if they were able to pick out the fine print that appeared along the bottom of the screen near the end saying “certain video content” was created by artificial intelligence.
When an announcer intones Talarico is “wrong for Texas” and Republican Senate nominee Ken Paxton says he approves the spot, they never say it’s AI.
Whether anyone was fooled by the ad and whether it was effective is hard to say. But analysts studying such “deepfakes” worry their use is already disturbingly deceptive and expect the prevalence of the digitally crafted messages will only grow, likely metastasizing in the 2028 presidential election after the technology has had two more years to improve.
Trending
What is clear is that campaign operatives already have found the lure of deepfakes irresistable, especially Republicans, and have deployed AI-generated forgeries in exponentially greater numbers than just two years ago.
According to a count by the Wesleyan Media Project, which monitors campaign advertising, the phony Talarico scene is one of more than 289 political ads using at least some AI imagery that have run in the 2026 election cycle as of Oct. 29, with more than $187 million spent to air them on TV, radio and digital platforms across the country.
Travis Ridout, a political science professor at Washington State University and a co-director of the Wesleyan project, said less than $50,000 was spent on such ads in 2024.
Republicans have been by far the most aggressive adopters of the artificial unreality, creating about 82% of the digital fakes running in races up and down the ballot, according to the Wesleyan data.
For example, voters in Wisconsin have seen videos of Democratic congressional candidate Rebecca Cooke, who is challenging Rep. Derrick Van Orden in a “toss up” race, on their social media feeds. They’ve seen an ersatz Cooke say she wants to “take advantage of as much corruption as possible” and “give every illegal alien free health care.”
In Pennsylvania, Gov. Josh Shapiro’s Republican challenger Stacy Garrity released computer-chip conjured portrayals of Shapiro discussing a sexual harassment scandal that once plagued his office and claiming credit for bringing now-reviled data centers to the state.
“Fact is, I was the one who fast-tracked data centers right here in PA,” the AI-generated Shapiro says in the ad, “right after they handed me $5 million in political contributions.”
Ridout said the increase in ads made using AI could be a sobering indication of what’s to come. It’s a cheaper way to make campaign videos, which can cost thousands of dollars, and the practice is only expected to spread into 2028 and beyond, he said.
“AI is going to be used in very creative ways,” Ridout said. “We’ll be seeing a lot more deepfakes. It will allow campaigns to launch the sort of attacks that they might not have been able to do before.”
One of the troubling aspects of the technological blandishments is it can be difficult to tell that the people in them are fake. Making matters worse, Ridout said, less than 30% of the ads have a disclaimer warning of their unreality.
“Most of those are in pretty small print. I’m not sure voters would even notice them,” Ridout said. “And even if you know what you’re watching is false because it’s labeled as false or synthetically generated or whatever, it still might stick in your mind when you’re going to vote, and that’s the real question right now.”
He expects the problem to get worse as deepfake use becomes more common. No clear rules governing their use exist yet in federal election canon, and states have taken only patchwork steps to regulate the phenomenon.
A handful of Democrats in the House have introduced legislation to curb AI in campaign ads and impose stricter guidelines when the use is undisclosed, signaling possible action if Democrats take control of Congress.
“As gaps in policy continue to allow campaigns, PACs, and scammers to test the limits of their deceit, action from Congress has grown long overdue,” Rep. Yvette Clarke (D-New York), who introduced a bill on the issue, told The Sun in a statement.
In the meantime, candidates who have been targeted by deepfakes are using a variety of tactics to respond.
Talarico hasn’t tried to have the advertisements pulled in Texas, where the law bans them within 30 days of the election date. Instead, he has used them in his messaging, airing TV ads that spotlight Paxton’s use of AI.
In one, Talarico walks through a warehouse of old televisions, saying he doesn’t need to use AI because “the truth about Paxton is bad enough.” Some of his campaign ads include the disclaimer: “Made by humans, not AI.”
“The most powerful people in the world know they can’t beat James Talarico with the truth, so they’re slandering him with AI deepfakes, foreign bots, and conspiracy theories to protect their broken, corrupt political system,” JT Ennis, a Talarico spokesman, said in a statement.
Attorneys for Cooke, on the other hand, filed a cease and desist letter that promised the campaign would “evaluate all legal options to address that misconduct.” A resident of the district also filed an ethics complaint with the Wisconsin Ethics Commission to try to halt the production of the social media videos.
“Politicians like Derrick are posting deep fakes instead of reversing the policies that sell us out. He can’t campaign on his record,” Cooke said in a statement. “Wisconsinites deserve a representative that is rooted in reality, and focused on solutions not distractions. These deepfake videos should be banned.”
In New York, Gov. Kathy Hochul, who has been similarly hit with artificial ads by Republican challenger Bruce Blakeman, has taken a similar approach. The New York State Democratic Party filed a complaint with the state’s board of elections in August, accusing Blakeman of violating state law.
“Team Hochul doesn’t need AI ads to tell voters about our opponent,” Ryan Radulovacki, a spokesperson for the Hochul campaign, told The Sun. “Bruce Fakeman has resorted to straight up lying to New Yorkers.”
The Republican campaigns declined to discuss their ads.
Ridout said the legal pathways for challenging fake political videos still remain largely untested, and it’s unclear whether there will be a viable path for candidates trying to pull them from circulation.
“Everyone’s figuring it out right now,” Ridout said. “A lot of these state laws are new, untested. A legal recourse often takes a long time and so the election’s already over by the time you get some action on something.”
If federal and state lawmakers want to make rules clear, they’ve got less than two years to get it done before the next election, when the old-fashioned mudslinging may end up more like a mountain of slop.
Authors
