AI already costs $187 million in political ads and nobody can prove they work
The Wesleyan Media Project counted 289 ads made or enhanced with artificial intelligence in the 2026 cycle and $187 million behind them, more than 3,000 times the 2024 figure. Nine in ten came from the Republican side, which lost the election.

The 2026 campaigns spent about $187 million promoting ads generated or enhanced with artificial intelligence, on television and online, according to the Wesleyan Media Project count that PolitiFact published on November 9 and the Poynter Institute circulated on the 11th. The jump is brutal: in the 2024 cycle the equivalent figure was roughly $50,000, more than 3,000 times less. The tracking identified close to 300 distinct ads; on WBUR's On Point program on November 6, project co-director Erika Franklin Fowler specified that the latest count was 289 unique spots and warned that the tally is «certainly an undercount» and will keep growing.
The partisan split is nearly unanimous. About 90% of the tracked ads came from Republican candidates or groups supporting them, according to PolitiFact. Samuel Woolley of the University of Pittsburgh put the Republican share of the roughly 200 videos his team followed at «80% or more». The Texas Senate race left the most visible example: there were ads with AI-generated imagery attacking state Rep. James Talarico, the Democratic nominee, who won anyway. In money terms, the $187 million is a minimal fraction of the total: AdImpact projected around $11.6 billion in political ad spending for 2025 and 2026.
Nobody has been able to show the spots moved votes. Measuring their influence is hard because voters received messages across many platforms at once and impressions are not yet quantified. Cayce Myers, a public relations professor at Virginia Tech, held that «AI content isn't necessarily persuading audiences» and that it mostly reinforces opinions voters already held. Travis Ridout, co-director of the Wesleyan Media Project, explained that campaigns held back because of «reputational risk» and warned of voter backlash. There is a basis for that fear: a 2025 Pew Research Center study found that 71% of Americans said they would like a candidate less if they learned AI had helped write a speech they had enjoyed.
The legal framework remains a patchwork. There is no federal regulation of AI in political advertising, as Fowler summed up, and 31 states regulate deepfakes in political messaging, according to the National Conference of State Legislatures. Christina Walker, a University of Georgia professor who studies AI and political information, said these ads are already «the new normal»; Tim Harper of the Center for Democracy and Technology described them as a new medium for the old genre of the political cartoon. The front that grew quietly was a different one: according to a Reuters report cited in the article, campaigns began drafting their messages so that AI chatbots would serve them to voters.
Before putting AI into your production line, separate two decisions: using it to produce faster and using it to deceive. The first lowers costs; the second costs you. In 2026 the side that ran 90% of the AI ads lost, and 71% of voters penalize a candidate when they discover the machine behind the message. And open a front almost nobody is covering: write your messages so chatbots repeat them, because that is where your electorate is already getting informed.
Summary written by the FOCUS editors based on the original story. Rights to the original content belong to its publisher.
FOCUS courtesy space


