Artificial intelligence is quietly reshaping campaign advertising this cycle, even as the data centers that power the technology emerge as a flashpoint in the midterm elections. A new analysis from the Wesleyan Media Project finds that at least 164 ads this year have been created or enhanced using AI, with roughly 80 percent of them coming from Republican candidates or allied groups.
The trend is visible in high-profile races. In Texas, Attorney General Ken Paxton's campaign released an ad depicting his Democratic opponent, James Talarico, scribbling in a Bible with crayons. Talarico fired back on social media, accusing Paxton's team of using AI to fabricate attacks. In Georgia, GOP Rep. Mike Collins ran an ad featuring cartoonish AI images of Democratic Sen. Jon Ossoff receiving spa treatments, drawing a sharp rebuke from Ossoff's camp.
Michael Franz, a professor at Bowdoin College and co-director of the Wesleyan Media Project, called the situation ironic. Candidates on both sides are campaigning against data centers in their districts while relying on the same infrastructure to generate their ads. "The data centers are what generate the AI stuff that people are using," Franz said. "They're relying in a sense on these data centers to generate these materials for their ads but at the same time they might be telling voters that they don't want data centers in their own constituency."
The use of AI in political ads is not limited to major races. A viral AI-altered image of Michigan GOP Senate candidate Mike Rogers circulated online, and a pro-Trump PAC produced a Paxton ad featuring an AI-generated Talarico in a dress. Supporters of the technology argue it is no different from parody or political cartoons, but experts worry about the line between humor and deception.
"These ads are on a continuum from obviously comical to indistinguishable from real life," said Mark Jones, a political science professor at Rice University. "The more lifelike they become, the more potentially pernicious they are."
Despite concerns, federal regulation has not kept pace. While 31 states have enacted laws targeting AI in politics, there are no federal guardrails. Franz predicts campaigns will continue to adopt the technology "readily" without clearer rules.
Daniel Schiff, co-director of the Governance and Responsible AI Lab at Purdue University, warned that even benign-looking AI content can influence voters. "It's not just about deceptive deepfakes," he said. "Positive, lighthearted synthetic media can also shape opinions."
Sen. Raphael Warnock (D-GA) recently spoke out after AI images falsely suggested he was battling health problems. In a video, he called the posts "bizarre" and "scary," noting that constituents and friends had reached out to wish him well. "As these tools become more powerful, it will get harder to distinguish fact from fiction," he said. "What if hours before an election someone created fake stories about candidates? How will voters know what's true?"
The debate over AI in campaigns unfolds against a broader fight over data center expansion. Recent warnings from AI industry insiders, including former OpenAI and Anthropic researcher Jacob Coxon, who said "people building AI earnestly believe that it could kill us all by the end of the decade," have fueled public anxiety. President Trump dismissed such fears, arguing that a "high IQ" president is sufficient protection.
Public opinion reflects this unease. A Politico survey found that roughly two-thirds of Americans believe AI could eventually destroy humanity, and a Marquette University Law School poll showed similar levels of concern. As the midterms approach, the clash between technological adoption and voter skepticism is likely to intensify, with AI ads remaining a contentious feature of the campaign landscape.
