AI ads flood campaign trail despite growing data center outrage, ‘doomsday’ warnings

Ads shaped by artificial intelligence are everywhere on the campaign trail, even as AI data centers become a major midterm issue and a potential liability for Republican candidates.

In the Texas Senate race, Republican state Attorney General Ken Paxton’s campaign put out a new ad earlier this month depicting his Democratic opponent James Talarico as scribbling in a Bible with crayon — even as Paxton appears to take a tougher stance on AI infrastructure.

“Have y’all seen these AI deepfakes they’re making of me? They don’t have any real attacks, so they’re firing up their data centers and making up fake ones,” Talarico said on social media.

And in Georgia last week, Republican Senate hopeful Rep. Mike Collins and Democratic Sen. Jon Ossoff clashed over an ad that included cartoonish, AI-generated images of the senator getting spa treatments.

The political ads, mostly pro-Republican, are among at least 164 that have been generated or enhanced by AI this year according to a new Wesleyan Media Project report — highlighting a midterm contrast between the rising use of the technology and growing voter concern about the rapid build-out of data centers .

Michael Franz, a government and legal studies professor at Bowdoin College and co-director of the Wesleyan Media Project, said there’s “a bit of an irony” given candidates in both parties are opposing data centers as many of their campaigns utilize the technology powered by the facilities.

“The data centers are what generate the AI stuff that people are using,” Franz explained. “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.”

Candidates up and down the ballot — as well as groups that support them — have turned to AI technology to help generate attack ads and snag viral social media posts as control of Congress hangs in the balance this fall. Per Franz, 80 percent of these are from Republican candidates or GOP groups.

An AI-altered image of Michigan Republican Mike Rogers went viral amid his Senate campaign, posted by a former GOP congressional aide. A pro-Trump PAC issued a pro-Paxton ad this spring featuring an AI-generated Talarico in a dress singing about transgender children. Republican Spencer Pratt appeared as the hero of a Gotham-esque Los Angeles in an ad this summer knocking Democratic Mayor Karen Bass.

Proponents argue that AI-generated content is an extension of existing political norms, akin to parody or political cartoons.

But the use of generative AI in politics has raised experts’ worries that the technology’s ability to produce realistic videos and audio — depicting imagined or exaggerated information — could contribute to an electoral landscape already fraught with mis- and dis-information.

The ads “are on this continuum of being the obviously comical to being, you know, indistinguishable from real life, and the more that they are lifelike and realistic, the more potentially pernicious the ads become,” said Mark Jones, a political science professor at Rice University in Texas.

And although 31 states have moved to crack down on the technology — according to the National Conference of State Legislatures — there still aren’t any guardrails at the federal level.

“It’s a brand-new world here,” Franz added, noting that he suspects campaigns will “readily adopt and use” artificial intelligence more and more without clear regulation.

Daniel Schiff, the co-director of the Governance and Responsible AI Lab at Purdue University, cautioned against being “restricted to only the idea that highly deceptive, persuasively deepfakes matter,” stressing that even positive, lighthearted synthetic media “can influence people.”

In Georgia last week, Sen. Raphael Warnock (D) had to speak out against AI images that falsely claimed he was battling health issues, among other posts he called “bizarre” and “scary.” He said that many constituents and even friends had reached out to wish him well, informed by the false claims.

“As these tools become increasingly powerful, it will get harder and harder to distinguish fact from fiction,” Warnock said in a video posted to social media. “And this has huge implications for our democracy. What if hours before an election someone created a page with AI-generated images, fake stories about the candidates? How will voters know what’s true and what’s not true?”

The use of artificial intelligence in political communications comes against the backdrop of a separate, raucous debate over how lawmakers should approach the rapidly developing technology and the expansion of data centers across the country.

Warnings from several current and former AI company employees have stoked concerns about the tech’s potential dangers in recent weeks.

After Jacob Coxon, a former researcher for both OpenAI and Anthropic, remarked that the “people building AI earnestly believe that it could kill us all by the end of the decade” in a lengthy social media thread, President Trump gave a hearty defense of his administration’s approach — claiming that the only protection needed is a “high IQ” president.

In a Politico survey released last week , roughly two-thirds of Americans said they believe that AI could someday destroy mankind, and a similar share said they would “stop working on it” if they were in charge. A Marquette University Law School poll also found that two-thirds of Americans think AI is bad for society , though roughly 7 in 10 respondents said they use AI themselves.

An NBC News poll released Monday also found that, though Republicans were more supportive of pro-data center candidates than independents or Democrats, nearly half of GOP voters said they’d be less likely to support a candidate who backed data center construction.

“Any use of AI in an ad probably does now have to be evaluated based on its pros and its cons,” Jones said, adding that candidates will have weigh their advertising goals and “any potential blowback” for AI use.

Amid frustration over data center expansion in Texas, Paxton — who is trailing Talarico in the latest Decision Desk HQ averages of their Senate race — has shifted into a stronger stance on AI infrastructure. Rogers in Michigan made a similar move , adding to signs that candidates are treading carefully.

“It’s also one of those issues where I think Republicans are able to sort of see the trend here, see the public opinion, and at least on the issue of data centers, you know, set aside use of AI,” Franz suggested. “But when it comes to sort of building one [data center] in your town or your county or whatever, it seems like it’s going to be a political fight in the coming years with a lot of elected officials sort of opposed.”

Experts say the average voter may not yet connect all of the dots between synthetic media, data centers and other facets of the AI boom — but that the growing anxieties could eventually come to a head and make its use a riskier bet for political figures.

“We have this array of reasons that people may be opposing data centers, and this also isn’t totally clear yet: Is it water or environment or noise? Is it sort of cost of living associated with electricity, or is it precarity and labor disruption? There’s some bundle of reasons that we’ve gotten bipartisan pushback on … data centers,” Schiff said.

“If the public starts to see AI in their personal lives or their work, doing all of these advanced things, changing jobs, making people feel more precarious about the future, and they actually sort of take these very seriously, it would be very interesting to see if it influences their voting behavior.”