When Rep. Thomas Massie raised the alarm about an artificial intelligence-generated attack ad that falsely shows him in a romantic scenario with two liberal Democratic stars, he warned that it could mislead and rattle older voters.
The ad shows Mr. Massie holding hands with Reps. Alexandria Ocasio-Cortez and Ilhan Omar and checking into a hotel, under the tagline “Thomas Massie caught in a throuple!”
“It reeks of desperation, but they’re hoping the older generation won’t realize it’s an AI-generated lie,” said Mr. Massie, who faces Kentucky voters Tuesday against Trump-backed primary challenger Ed Gallrein.
Researchers who study aging and deception say Mr. Massie is onto something.
Natalie Ebner and Didem Pehlivanoglu at the University of Florida say older adults face a unique set of challenges when confronted with AI-generated photos, videos and audio that look uncannily real. The Massie ad, funded by the MAGA KY super PAC, is a prime example.
“This is worse than adultery,” a narrator says in the spot. “It’s a complete and total betrayal of President Trump and Kentucky conservatives.”
A small disclaimer notes that the spot was created with artificial intelligence, but the visuals do the work.
Ms. Ebner and Ms. Pehlivanoglu say that is exactly the danger.
AI deepfakes often have a smooth, polished “hyper-realism” that the aging brain can mistake for authenticity. It makes them easier for the brain to process — and that ease can be misleading.
“They basically look more real than real material,” Ms. Ebner said. “The simpler the information, the easier it is on the aging brain.”
Layer on “truth bias” — the instinct to assume what we see is real — and the risk grows. That bias tends to strengthen with age and becomes even more powerful through repetition. As Ms. Pehlivanoglu put it, “As it repeats, the brain confirms: Oh, I think this is real.”
The political world is adjusting to this new terrain, which will serve as part of the backdrop to the Kentucky primary, when voters will decide whether they still want Mr. Massie despite his running afoul of President Trump.
David Martin, co-founder of the ad-tech firm Adwave, said the rise of AI has blown apart the old cost barriers around television advertising.
What once required an agency, a production studio and media buyers — a process that ran $10,000 to $15,000 — can now be done for as little as $50, with a finished spot ready to air the next morning.
That shift has opened the door not just for major campaigns but also for mayors, sheriffs and other local candidates who never had a shot at television before.
Because TV still commands a level of trust that social media never has, older viewers who encounter a deceptive AI-generated ad on a familiar network may be even less likely to question what they are watching.
“You have these periods in time, especially with technology, with new technology, emerging technology, where it’s kind of like the wild, wild west,” Mr. Martin said. “Right now, we are absolutely in that time with AI. The technology is progressing faster than our legislation and our laws, frankly, are progressing, and there’s very little regulation out there.
“It’s almost lawless, which is kind of frightening,” he added.
A proliferation of deepfake ads
Taken together, the collapsing cost of AI ads, the particular vulnerabilities of older voters and the partisan incentives to provoke outrage have created a perfect environment for this kind of content to spread.
That shift is already visible.
Former reality TV star Spencer Pratt has jolted the Los Angeles mayor’s race with a string of AI-generated ads, including a Batman-style spot casting him as a caped crusader trying to save the city from Mayor Karen Bass, who is running for reelection and appears as a Joker-like foil.
The experimentation is not limited to California. In Texas, Attorney General Ken Paxton blasted Sen. John Cornyn with an AI-generated ad showing him dancing with Rep. Jasmine Crockett, a liberal Democrat who infuriates the Republican base. Mr. Cornyn’s team hit back with a B-52s-themed spot featuring Mr. Paxton partying with alleged mistresses at his “Love Shack.”
National Republicans are joining the fray. The National Republican Senatorial Committee, the Senate Republicans’ campaign arm, aired an AI-generated ad depicting Democratic nominee James Talarico reading his old social media posts, including a 2021 line where he wrote, “Radicalized White men are the greatest domestic terrorist threat in our country.”
In Maine, presumptive Democratic nominee Graham Platner has been depicted as a shirtless buffoon behind a computer.
For most Americans, this is all brand-new territory.
A 2023 Pew survey found that nearly 6 in 10 Americans are not sure what a deepfake even is. Researchers at Princeton University and New York University’s Social Media and Political Participation Lab found that during the 2016 presidential election, less than 9% of Americans shared links to “fake news” sites on Facebook, but the behavior was far more common among people 65 and older.
Voters 65 and older shared roughly seven times more false information on Facebook than voters ages 18 to 29.
Still, the researchers stress that these vulnerabilities among older adults are not universal. Older adults who are more analytical or more tech-savvy can spot manipulated content.
Partisanship can play a role
Others argue that cognitive decline is not the sole reason why older voters are more willing to engage with false content and pass it along.
Guilherme Ramos, an assistant professor at the Rochester Institute of Technology who co-authored a study with researchers at the University of Colorado Boulder, found that adults 55 and older were likely to share more misinformation because they are more partisan.
“While everyone tends to believe in and share information that benefits their own group, older adults display these partisan tendencies even more,” he said of his study, which focused on written news stories.
Asked about the “throuple” ad, he said, “I think the danger is not just that older voters do not know AI exists; it’s that AI makes these false stories more vivid and emotionally ‘sticky.’”
“This can amplify believability and increase the likelihood of sharing the news if it aligns with a person’s partisan narrative,” he said.
Taken together, the research suggests older voters may face a double exposure: They might be more likely to be fooled by a convincing deepfake and more likely to pass it along once they believe it.
Mr. Ramos said that dynamic is exactly what makes AI-generated content so potent.
“Even before this technology, people shared false narratives or out-of-context images to achieve that same dual goal,” he said. “What I think is new is that these AI-generated videos make everything so vivid, believable and engaging that people might be even more likely to share them.
“I do think that these AI ads serve that dual goal, and perhaps they do so more effectively — by amplifying sharing tendencies — and more efficiently, since AI content is easier and cheaper to produce than traditional media.”
