Campaigns and PACs have spent more than $137 million on AI generated political ads ahead of the 2026 midterms, according to a Wesleyan Media Project report covered by Fortune on October 21. In the 2024 cycle, the same researchers detected less than $50,000. More than 70% of the 220 AI ads they found carry no AI disclosure. For anyone who buys media, this is the clearest signal yet that synthetic creative has gone mainstream.
What the Wesleyan Media Project found
The Wesleyan Media Project has tracked political advertising for years, and this count runs through October 11. It identified 220 campaign videos made at least partly with AI, backed by more than $137 million in spend. Two years ago that figure was a rounding error.
"We went from about zero in spending on AI ads in the last election cycle," said Travis Ridout, the project's codirector and a political science professor at Washington State University.
The split is lopsided. Republican campaigns and PACs account for 81% of the AI ads and 88% of the spending. Ridout ties that to the party's warmer view of AI overall.
The examples show the range. In the Texas Senate race, a Ken Paxton ad shows an AI version of James Talarico writing in a Bible with a crayon and delivering a monologue he never gave. Talarico's response: "They are lying about a candidate in this race." In Pennsylvania, Stacy Garrity's campaign ran an AI ad depicting Josh Shapiro praising AI data centers, and that one did disclose its origins. Most did not. More than 70% of the AI videos ran with no label at all.
What this means for marketers
Political campaigns are the most deadline driven advertisers in the country. They spend fast, cut creative overnight, and live or die by turnaround. When they move $137 million into AI production in a single cycle, commercial advertisers should read it as proof of the economics, not a novelty.
Ridout made that point himself: AI lets lower budget candidates produce ads they could never afford before. Swap "candidate" for "local business" and the logic holds. A dentist, a roofer, or a regional ecommerce brand can now test ten video concepts for what one shoot used to cost.
The disclosure number is the warning. Seven in ten of these ads ran unlabeled, and Ridout says the result is voters who feel "I don't know what to believe anymore." Political speech gets wide legal latitude. Commercial advertising does not. Brands answer to the FTC's rules on deceptive ads, and Meta and Google already require AI disclosure on political and issue ads that use synthetic people or scenes. A complaint costs far more than a disclosure line.
The smart play is to use AI for speed and volume, then label it plainly. Tools like KenjiAI (kenjiai.com) are built for exactly this kind of shift, letting small teams produce and test AI creative at a pace that used to require an agency. The output still needs a human deciding what is true and what gets disclosed.
What to watch next
- Disclosure laws. Talarico has pushed legislation to ban AI deepfakes, and a growing number of states already regulate synthetic political media. Expect that pressure to reach commercial ads once the midterms are over.
- Platform enforcement. If 70% of these ads ran without labels, either the platforms missed them or their rules have gaps. Watch for policy changes from Meta, Google, and YouTube after November.
- Your own creative pipeline. Audit every AI generated ad you run today. If a reasonable viewer could mistake a synthetic person or scene for real footage, add a label before a competitor or a regulator points it out.
Campaigns proved the cost math in one cycle. The $137 million figure will look small by 2028, and the brands that build disclosure habits now won't have to retrofit them under pressure.
Source: Fortune, October 21, 2026, reporting on the Wesleyan Media Project.
Published by the Media Traffics | KenjiAI team. kenjiai.com
Top comments (0)