Scroll long enough this fall and you might see Texas Senate candidate James Talarico in a dress, singing a parody of “My Favorite Things.” Or Kentucky Rep. Thomas Massie, apparently dining and laughing with two of his Democratic colleagues in a romantic relationship. Neither scene ever happened. Both were ads, built with artificial intelligence and aired to millions of voters.
At first glance, this explosion of political advertising may look like a good thing. If candidates are spending billions telling voters what they believe, shouldn’t voters be more informed? But that simply isn’t the case. Combined with easily generated AI content, it seems to have created the perfect storm: a cycle of political advertising that seems impossible to escape.
Candidates are competing for our attention on social media, streaming services, billboards and just about every other screen we look at. Political advertising has become so cheap, constant and optimized for attention that campaigns are increasingly incentivized to win a battle of attention rather than actually explain policies. The problem with this constant stream of political messaging is that, eventually, people stop listening.
The numbers are hard to ignore. A recent analysis by The Washington Post of more than 8,000 advertisements in competitive national and statewide races found roughly $1.6 billion in spending. References to the state of the economy appeared in nearly three out of every five ads, along with concerns regarding housing, grocery prices, gas and health care. When AI is thrown into the mix, the issue gets even more complicated. In an interview with Fox News, Democratic strategist Chris Moyer acknowledged that AI is making powerful campaign tools cheap and easy to obtain.
Consider the ad the pro-Trump PAC MAGA Kentucky ran against Rep. Thomas Massie in the Republican primary for Kentucky’s 4th District. The so-called “throuple” ad used AI to show Massie in a romance with two Democratic colleagues, with realistic-looking footage of him walking, dining and laughing with them. The ad opened by saying it was satirical and AI-made, but the images look real enough that viewers could easily take away something that never happened. Some other ads this cycle haven’t disclosed their AI use at all.
The problem isn’t that political advertising exists. Political ads have existed almost as long as our democratic processes have, with pamphlets and broadsides being used in the earliest presidential elections. When used responsibly, political ads can introduce voters to candidates, explain issues and encourage civic engagement.
The problem lies in what occurs when every candidate, across every level of government, is fighting for the same seconds of your attention span.
Instead of slowing down and explaining complicated issues, campaigns have an incentive to make their messages short, emotional and memorable. Ads are designed to utilize sound bites and half-truths to capture your attention – even if it is only for a second. A candidate’s position on a complicated economic issue is harder to fit into a 15-second TikTok than a simple attack against an opponent.
When voters see dozens of these messages every day, they can to blur together. This creates a strange contradiction. We have access to more political information than ever, yet political advertising can make it harder to figure out what candidates actually want to do.
There is also another complication: AI. Researchers have tracked millions of dollars in spending on AI-generated political advertisements during the 2026 election cycle. As technology makes it easier and cheaper to create political content, the amount of messaging voters encounter could continue to grow. Worst of all, there is no federal regulation for how AI is allowed to be used in political ads, or whether ads need to disclose AI use. Right now, AI ad regulation is a state or locally controlled issue.
One of the many examples of this can be seen in the Texas Senate race, where Republican candidate and State Attorney General Ken Paxton released an AI-generated ad targeting his Democratic opponent, State Rep. James Talarico. Paxton’s ad featured an AI-generated Talarico reciting statements on religion and gender that are mostly based on things he really said, but reworded, stitched together and stripped of context. For example, his “Christian who hates Christianity” line came from a podcast where he explained why he keeps returning to Christianity, and the “six genders” line distorts a 2021 remark about six biological sexes, which he has since walked back. A tiny-print AI disclaimer appears partway through. The Talarico campaign called the ad a “deepfake.”
The crux of this issue is that more advertising does not automatically mean more informed voters.
Political campaigns should still advertise. Candidates need ways to reach people, especially voters who may not know much about a particular race. But there is a difference between informing voters and overwhelming them.
The goal of political advertising should not simply be to make voters remember a candidate’s name or dislike an opponent. It should give voters enough information to ask a much more important question: What would this person actually do with my vote?
That question deserves more than a 30-second attack ad. Our system relies on voters weighing ideas to make decisions, but when half-truths or even whole truths are squeezed into 10-second TikTok ads, politicians miss the point.
