Loading page…
The new count shows synthetic media is becoming routine campaign material, while rules for political AI remain fragmented across the states.
Listen to this story
A Wesleyan Media Project review identified 164 AI-generated or AI-enhanced political ads during 2026, with roughly four in five tied to Republican candidates or groups. The dataset mixes obvious caricature with realistic or misleading synthetic imagery, making the category broad but highlighting a growing campaign tactic. Thirty-one states have enacted some rules, while federal safeguards are absent.
The tally includes both fully generated and AI-edited material, from cartoon attacks to realistic-looking false claims.
Examples include AI depictions involving Ken Paxton, James Talarico, Mike Collins, Jon Ossoff, and Raphael Warnock.
Mark Jones of Rice University said potential harm increases as synthetic content becomes harder to distinguish from reality.
Supporters describe AI campaign material as another form of parody or political cartoon. But the Wesleyan Media Project has identified at least 164 AI-generated or AI-enhanced political ads in 2026, with about 80% coming from Republican candidates or GOP groups.
The count covers both fully generated and AI-enhanced ads, a category broad enough to include deliberately cartoonish attacks and more realistic-looking synthetic media. Mark Jones, a Rice University political science professor, warned that the potential for harm rises as the material becomes more lifelike and harder to distinguish from reality.
The Wesleyan Media Project identified at least 164 AI-generated or AI-enhanced political ads in 2026.
Project co-director Michael Franz said about 80% of the identified ads came from Republican candidates or GOP groups.
Ken Paxton’s Texas Senate campaign released an AI-related ad depicting Democratic opponent James Talarico scribbling in a Bible with a crayon. In Georgia, an ad associated with Republican Senate candidate Mike Collins used cartoonish AI-generated images of Democratic Sen. Jon Ossoff receiving spa treatments.
Not all of the examples are presented as satire. Sen. Raphael Warnock publicly denied AI-generated images falsely claiming he had health problems, saying constituents and friends had contacted him after seeing the claims. The episode shows how synthetic political imagery can reach people beyond the original post or ad.
Thirty-one states have taken steps to regulate AI in political communications, according to the National Conference of State Legislatures, but there are no federal guardrails. The contrast leaves campaigns operating under different rules even as AI-made and AI-edited material spreads through competitive races.
Story updates
AI-generated campaign ads have substantial money behind them, but most identified examples carry no AI label. In an October 8 WSHU story, Wesleyan Media Project co-director Erika Franklin Fowler warned that limited disclosure leaves campaign use undercounted. The project counted 164 ads through early September 2026, backed by at least $80 million in spending.
Fowler, also a government professor at Wesleyan University, expects the number to accelerate in the weeks before the midterm elections.
The Connecticut-based, nonpartisan project has tracked and analyzed political advertising for 16 years. Fowler said its researchers identified AI use in ads from both parties, in races across the country. The findings cover a range of generated depictions, rather than one campaign, party or style of political messaging.
The project’s initial May report identified 33 AI-generated political ads with about $20 million in spending. By early September, the identified total had reached 164, with at least $80 million behind those ads. Those figures describe spending backing the ads, not simply the number of generated images or candidate depictions.
A later estimate suggests the tally continued to grow. In the Washington Post on September 30, project co-director Michael Franz estimated that the number of ads using AI was approaching 180.
It’s a multipronged system but it's also really important to say that this is absolutely an undercount of AI used in campaigns because in a lot of ways there’s little disclosure
Almost 70% of the identified ads lacked a disclaimer disclosing AI use. Ads without those disclaimers accounted for 78% of the tracked spending.
Not all the identified material serves the same purpose. Fowler described satirical portrayals intended to poke fun at candidates, alongside hyperrealistic deepfakes—AI-generated depictions made to look real. Those realistic portrayals can appear in attack ads or make a candidate look more favorable.
Fowler pointed to the absence of a federal disclosure requirement. Some states require disclaimers for political ads containing AI-generated content, while some state laws prohibit deepfakes. Disclosure rules and outright restrictions are different responses: one identifies synthetic content for viewers, while the other limits its use.
Connecticut and New York illustrate that uneven response. Both introduced bills in 2025 seeking to ban the creation of election-related deepfakes within a specified period before an election. New York’s legislation became law. Connecticut passed other AI-related bills, but did not enact an election-deepfake restriction.
washingtonpost.comLoading discussion...
Join the conversation
Explain which kinds of ads need disclosure.
Be the first to share a perspective or an experience.
Reader comments
Newest comments first. Replies stay oldest first.