Campaigns and outside groups have spent more than $137 million on political ads made at least partly with AI ahead of the 2026 midterms, according to Wesleyan Media Project research first reported by The New York Times on October 16. In the 2024 cycle, researchers found less than $50,000. Only 27% of these ads disclose the AI. If you buy ads for a living, this is the clearest preview yet of where synthetic creative is heading, and how audiences will react.
What the Wesleyan Data Shows
The Wesleyan Media Project, co-directed by Erika Franklin Fowler and Travis N. Ridout, identified 220 AI-made political ads running across 34 states. The researchers say the true count is likely higher, because they can only flag ads they can verify.
The jump in money is the headline. Under $50,000 two years ago. More than $137 million now. That is roughly a 2,700x increase in a single election cycle.
The disclosure numbers are worse. Gizmodo's breakdown of the data found just 27% of the ads carry any label saying AI was involved, even though 30 states now require AI disclosures on political ads. Nearly half of the ads feature deepfake depictions of real politicians.
The spending is lopsided. Republicans account for 81% of identified ads and 88% of the spend, with Democrats at 12% of ads. Bruce Blakeman, the Republican candidate for New York governor, has reportedly posted at least 22 AI-generated ads since April, and the New York State Democratic Party filed a complaint with the state Board of Elections in August. His campaign argues satire needs no disclosure. On the other side, Texas Senate candidate James Talarico ends one ad with a line that tells you where this is going: "Made by humans. Not AI."
What Marketers Should Take From This
Political ads are a stress test. Campaigns have short deadlines, huge volume needs, and little patience for slow creative teams. When AI creative goes from a rounding error to nine figures in one cycle there, commercial advertisers are watching the same economics play out.
Two lessons stand out.
- Production cost is no longer the bottleneck. A campaign can now test dozens of video concepts for the price one used to cost. Brands that still brief one hero spot per quarter will get outspent on variation alone.
- Trust is the new bottleneck. An Auros study from this month found 44% of U.S. shoppers say they avoid brands that use AI in ads. Talarico's "Made by humans" tag works because voters are primed to suspect everything.
The smart play sits between those two facts. Use AI for speed and testing volume, and stay upfront about it. Tools like KenjiAI (kenjiai.com) are built for exactly this kind of shift, handling the repetitive production and follow-up work so a small team can run the volume of a big one while real people still own the message and the relationship.
The campaigns hiding their AI use are borrowing against their own credibility. Some of them will get caught before November 3, and the clips of them getting caught will travel further than the original ads.
What to Watch Next
Three things matter over the next few weeks.
- Enforcement. With 30 states on the books and a 27% disclosure rate, the first fines or takedowns will set the tone. Watch the Blakeman complaint in New York closely.
- Platform policy. Meta, Google, and YouTube already require political advertisers to flag synthetic content. Expect tighter rules after the election, and expect them to creep into commercial ad policies next.
- Your own disclosure policy. Write one now, before a regulator or a viral screenshot writes it for you. Decide which AI uses you will label, how, and who signs off.
The midterms are about two weeks out. Whatever happens at the polls, the $137 million figure is going to show up in every agency pitch and every brand safety meeting for the next year. Know the number, and know where you stand on it.
Source: Wesleyan Media Project data, as reported by The New York Times (October 16, 2026) and Gizmodo.
Published by the Media Traffics | KenjiAI team. kenjiai.com
Top comments (0)