Politics / Technology

Add as a preferred source on Google

AI-Assisted Political Ads Draw More Than $137 Million Ahead of U.S. Midterms

Wesleyan counted 220 ads, with Republican-affiliated sponsors accounting for 88% of estimated spending. A federal filing lists $92,187.50 in expenditures opposing Michigan Senate candidate Abdul El-Sayed.

Updated 2 min read
A passerby watches a campaign video on a phone outside Michigan’s Capitol in Lansing. (AI-generated image / EconoTimes)
A passerby watches a campaign video on a phone outside Michigan’s Capitol in Lansing. (AI-generated image / EconoTimes)

The Wesleyan Media Project estimates more than $137 million has gone to AI-assisted political ads in the 2025-26 U.S. election cycle as the Nov. 3 midterms approach. Its Oct. 15 update counted 220 ads that used AI in at least part of their production and compared the spending with less than $50,000 identified for AI-generated ads in 2024.

AdImpact supplied the expenditure estimates, which Wesleyan analyzed. The project warned that some ads may escape its count when AI involvement is difficult to detect.

Republican candidates or groups sponsored 81% of the identified ads, compared with 12% for Democrats. Republican-affiliated sponsors represented 88% of estimated spending. The ads appeared in 34 states, and 27% included an AI disclaimer.

One example is a video from The Front Line Action targeting Michigan Senate candidate Abdul El-Sayed. Its AI-generated scenes show cards displaying “Death to America” and “Celebrate 9/11”; the video discloses that AI was used.

A Sept. 14 Federal Election Commission (FEC) filing lists two independent expenditures opposing El-Sayed: $80,000 for media and $12,187.50 for media production. It records Sept. 12 as the public distribution date but does not identify a specific ad.

An AAPC Foundation survey of 1,568 likely 2026 general-election voters, conducted Sept. 14-22, found strong opposition to fabricated videos depicting political opponents. Sixty-six percent of respondents were very concerned deepfakes could change how other people vote, while 33% had the same level of concern about their own vote.

NYU’s Center on Technology Policy distinguishes deceptive deepfakes from AI political material that makes its artificial origin apparent, a category its researchers call “political slopaganda.” The report says current evidence does not establish whether this content changes election outcomes. Researchers suggest campaigns may find it more useful for mobilizing voters or raising money than for changing candidate preferences. They also identify possible risks to public trust and partisan hostility, without presenting those as measured effects.

In 2024, the FEC declined to open a new rulemaking on AI campaign ads. It said existing federal law against fraudulent misrepresentation applies to AI-assisted media when the law’s requirements are met. The Federal Communications Commission (FCC) separately proposed disclosure requirements for certain radio and television political ads.

The Nov. 3 midterms will decide every U.S. House seat and roughly one-third of Senate seats, putting control of both chambers at stake.

Related articles

View all