AI-generated political – From Oklahoma to Texas to Los Angeles, the rush to generate realistic political visuals is colliding with a simple question voters can’t afford to ignore: if AI content isn’t labeled, what are people actually seeing?
Last Friday in deep-red Oklahoma, residents may have caught a glimpse of a “lifelong Republican” smiling at them, with a pin on his chest featuring Barack Obama.
A week later in California, viewers saw three men in a sunny backyard, sipping beers and repeating the same line as they explained why they supported a Republican mayoral candidate. “I’m not MAGA or anything,” they said.
None of it happened that way. These scenes appeared in content promoting candidates in competitive elections. And none of the images included text identifying them as having been generated by artificial intelligence.
The concern isn’t limited to obviously fabricated clips. Recent technological improvements have made images easier and cheaper to create—some can be recognized immediately as computer-generated. while others may look “passable as realistic to a casual observer. ” like the video of the three unidentified men in California. Some, though, have shown up as something closer to photo-real persuasion, including an ad attacking Oklahoma gubernatorial candidate Charles McCall. There was also a paid mailer sent by Sen. John Cornyn’s campaign this week in Texas, where he is in a run-off against a Trump-backed MAGA challenger.
That mailer shows the challenger, Ken Paxton, cavorting with well-known Democrats. Paxton is shown eating ice cream with Rep. Alexandria Ocasio-Cortez. riding bicycles with Senate Minority Leader Chuck Schumer. playing basketball with Democratic senate nominee James Talarico. and listening to House Speaker Nancy Pelosi—who. in the mailer. is holding a teacup.
The Cornyn campaign declined to comment.
For longtime political ad makers, the speed and realism are what make the problem feel urgent. “I would argue we’re just past the point where AI does generate indistinguishable content. ” said Rick Wilson. a longtime admaker and co-founder of the anti-Trump group The Lincoln Project. “It kind of freaks me out a bit, I’m not going to lie,” Wilson added in an interview.
Wilson said AI is also changing how quickly ads can be produced. “The speed of ads was always constrained by finding human talent,” he said. Now, he argued, an ad could be made using a prompt like “generate a 45-year-old woman facing economic pressure.”
If that sounds like a technical shift. the political effects have already begun to surface—most visibly in the Los Angeles mayoral race. There. the strongest challenger to Democratic incumbent is a Republican whose first-time candidacy is being promoted by AI-generated videos that have drawn millions of views online.
One video featuring a multi-ethnic group of women at a pilates studio—each saying they support the Republican candidate. Spencer Pratt—has 1.9 million views on X. The video of the three men has more than 2.2 million views. A Star Wars-themed video has 3 million views. Another video shows Pratt as Batman leading a tomato-throwing crowd against a Joker-painted Mayor Karen Bass and regal-looking Gov. Gavin Newsom; it has more than 5 million views. And those figures reflect views “just in May.”.
The effectiveness of AI in ads isn’t guaranteed in every case, and it can cut both ways. One study last year found that “genAI-created ads consistently outperform both human-and genAI-modified ads. increasing click-through rates by up to 19% in field settings.” But the same study also found that “disclosing AI involvement in ad generation significantly reduces advertising effectiveness by up to 31.5%. underscoring trade-offs relevant to evolving AI disclosure policies.”.
Right now, there is no across-the-board requirement that campaign ads label AI-generated content. A report from the Congressional Research Service from Sept. 2025 said there is currently no mandate requiring campaign ads to “alert the audience. or regulators. to the presence of AI-generated content.” That report also noted that the American Association of Political Consultants issued a statement in 2023 condemning “deceptive generative AI content in political campaigns.”.
Lawmakers have introduced bills aimed at forcing disclosure. In 2023, Rep. Yvette Clarke, D-New York, introduced legislation; last year, Sen. Amy Klobuchar, D-Minn. Both measures were referred to committee and did not come up for a vote.
Even if disclosure rules eventually arrive, the problem may not disappear. Professor Michelle Nelson. who teaches media literacy and campaign advertising at the University of Illinois Urbana-Champaign. warned that voters may not hold onto the fine print. She told Straight Arrow that people may remember an image. but. over time. they may forget what disclosures may have come with it. Nelson described the “sleeper effect. ” in which the impact of a negative message increases over time as details that might soften the attack are forgotten. “Our memory,” Nelson said, “is faulty.”.
In the space between speed and trust, the campaigns are moving faster than the rules. Oklahoma residents saw a “lifelong Republican” wearing an Obama pin; Texans received a mailer picturing Ken Paxton among top Democrats; and Los Angeles watched AI-amplified videos rack up millions of views—without any labeling to tell viewers what was real and what was generated. For voters. the result is a new kind of uncertainty: not just whether a campaign is persuasive. but whether it’s telling the truth about what they’re showing.
AI campaign ads political misinformation Oklahoma gubernatorial Charles McCall John Cornyn Ken Paxton Alexandria Ocasio-Cortez Chuck Schumer James Talarico Nancy Pelosi Los Angeles mayoral race Spencer Pratt Karen Bass Gavin Newsom media literacy