Political advertisers have poured more than $137 million into AI-generated ads ahead of the 2026 midterm elections, according to a new analysis from the Wesleyan Media Project, a sharp escalation that underscores how quickly generative tools have moved from novelty to standard campaign equipment. Researchers tallied 220 political ads this cycle that were generated at least in part by AI tools, a figure the analysis itself flags as almost certainly an undercount, since AI use is not always detectable and is frequently not disclosed.
The spending figure is a new high-water mark for AI-made political advertising. During the 2024 cycle, the Wesleyan Media Project identified less than $50,000 in spending on such ads. Two years later, the total has swelled by a factor of thousands, and AI-generated spots have aired across 34 states. For readers tracking the latest AI developments, the midterms are shaping up as the first federal election cycle in which generative AI is a mainstream production tool rather than an experiment.
What the Analysis Found
The Wesleyan Media Project, which tracks political advertising from Wesleyan University, counted 220 ads this cycle that were generated, at least in part, by AI tools. The $137 million attached to those placements is an eye-popping sum by any measure, and researchers caution the real total is likely higher because campaigns are not required to disclose AI use in most jurisdictions.
The partisan breakdown of the identified ads is lopsided. Of the 220 AI-generated spots, 81 percent came from Republican advertisers, while just 12 percent were paid for by Democrats. The remainder came from independent candidates or from groups advertising on ostensibly non-partisan issues.
The Disclosure Gap
Only 27 percent of the AI-generated ads identified in the analysis contain a disclaimer disclosing that AI was used. That gap matters because the legal landscape remains a patchwork: just 30 states require some sort of disclaimer when AI content appears in political ads, leaving voters in the other 20 states with no guarantee they will ever know a spot was machine-made.
The New York Times, which reported the $137 million figure on Thursday, noted that the adoption has outpaced the rules governing it. Federal legislation on AI disclosure in political advertising has stalled, leaving states to experiment with their own requirements of varying scope and strength.
Why Campaigns Are All In
The economics and speed of generative tools explain much of the adoption. A survey released earlier this year by the American Association of Political Consultants found that 83 percent of political professionals now use AI tools for work at least once a week, up from 59 percent the previous year. Nearly half of those surveyed said they use AI tools multiple times a day.
Campaigns and ad-makers deploy the technology across the production pipeline: drafting and refining ad scripts, cutting voiceovers, analyzing data, and writing fundraising emails. Under the compressed deadlines of an election calendar, the ability to spin up dozens of ad variants targeting narrow audiences has made AI tools an enticing option for consultants of both parties.
The AAPC survey also revealed a partisan divide among practitioners, with more Republicans reporting daily AI use than Democrats, a pattern that mirrors the 81-to-12 split in AI-generated ad placements found by the Wesleyan Media Project.
Voters Are Wary of the Worst Uses
While campaigns race to adopt the tools, the people they are trying to persuade remain deeply skeptical of the most manipulative applications. A survey released last week by the AAPC Foundation found that 95 percent of likely 2026 voters said it is unacceptable for campaigns to use AI to create a video of an opponent saying something they never said.
That tension, between widespread behind-the-scenes adoption and broad public discomfort with deceptive uses, is likely to define the next phase of the debate. Regulators in several states have already moved against AI deepfakes in campaign material, and the steady climb in AI-generated ad spending gives those efforts fresh urgency heading into the final weeks of the cycle.
What to Watch Between Now and Election Day
Three numbers are worth tracking. The first is spending: at $137 million with weeks of advertising still to run, AI-generated ad totals will keep climbing. The second is disclosure: if the 27 percent disclaimer rate holds as spending rises, the majority of AI-made ads will reach voters unlabeled. The third is geographic: ads have already run across 34 states, but only 30 require any disclosure at all, a mismatch that keeps the incentive to stay quiet firmly in place.
The midterms were already expected to be the most AI-saturated elections in US history. The Wesleyan Media Project's tally suggests that prediction has arrived early, and that the industry's disclosure practices have not kept pace with its production habits.
Researchers themselves treat the 220-ad count as a floor rather than a ceiling. Because AI-generated content is not always easy to detect, and because campaigns are under no obligation to volunteer it, much of this cycle's AI-assisted output may simply be indistinguishable from conventionally produced ads, visible only in the aggregate spending figures rather than in any labeled spot. That measurement problem is its own finding: two years after AI political advertising registered as a rounding error, it has grown large enough that nobody can say precisely how big it is.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
