Real race, fake videos: AI ads take over the Texas Senate fight
Oct. 30, 2026 · AI news · About a 4-minute read
Texas's U.S. Senate race between Republican Attorney General Ken Paxton and Democratic state Rep. James Talarico has become the country's biggest test of AI-made campaign ads. Polls show a close race heading into Tuesday's election, with Talarico holding small leads in most recent surveys.
This is a closer look at one race. For the national numbers, $137 million in AI ads and who's paying, read The first AI election. To check a clip yourself in 30 seconds, see How to spot an AI political ad.
What's happening
- Paxton's campaign and groups backing him, including the National Republican Senatorial Committee, have run ads showing an AI-generated Talarico doing and saying things he never did on camera. Examples include writing in a "King James Talarico Bible" and reading his old social media posts with made-up reactions.
- Many carry an "AI generated" label, though in some ads it is small or only shows for a few seconds.
- Paxton also used AI versions of fellow Republican John Cornyn and Democrat Jasmine Crockett during the primaries.
- Talarico has sworn off AI. One of his ads shows him walking past rows of TVs playing the AI attack ads, saying he doesn't need AI to make the case against Paxton. Another ends with the tag "Made by humans, not AI."
- Nationally, Republican candidates and groups account for most AI ad spending this cycle. Our $137 million story has the full breakdown.
On the one hand: the case for the ads
- Paxton's campaign says the ad shows a fictionalized Talarico and that some of the lines are based on things he actually said. One example is a 2021 remark in which Talarico described God as "nonbinary."
- Many of the ads are labeled as AI. Researchers at NYU found that most AI ads so far this year either used AI to polish an ad, such as replacing stock footage, or were obvious over-the-top satire rather than convincing fakes.
- Political parody is as old as politics itself, from cartoons to late-night sketches, and it gets strong free speech protection. Legal analysts have long questioned whether outright bans on political deepfakes would survive a First Amendment challenge.
- Supporters see AI ads as a cheap, fast way to grab attention and make a point about an opponent's record.
On the other hand: the case against
- Talarico says the ads put words in his mouth and are meant to mislead voters. A PolitiFact check found his critics, including Paxton, often left out part of one of his quotes about transgender kids, changing its meaning.
- Critics say small or brief AI labels are easy to miss, especially on a phone or TV.
- Democratic strategists are betting voters care more about authenticity and cost of living than flashy video, and some research suggests AI ads can backfire on the candidate they're meant to help.
- Neither party has a spotless record. Crockett's Democratic primary campaign wouldn't say directly whether a big-crowd image in one of its ads was AI-made.
The law, in brief
We covered Texas's 30-day deepfake law in our earlier story. The key point for this race: that law covers state races, not federal ones like this Senate seat, and there is no federal law on AI political ads.
The law is already being tested in a state race. Democratic lieutenant governor nominee Vikki Goodwin filed a criminal complaint against Lt. Gov. Dan Patrick over two AI videos of her. Patrick's campaign called them obvious parody and said they were posted more than 30 days before the election. The videos have since been taken down.
My take
The tools making these ads are the same kinds of AI video tools I review here, and they're only getting better. Whichever way you vote, the lesson is the same: if a political clip looks wild, pause before you share it. Look for an AI label, find the full original quote, and check the candidate's own channels. My how-to walks you through it. Fake or not, the vote you cast is real.
Read next
From MKUltra to AI: the long history of messing with your mind. Science Bits on how propaganda went from war posters to AI bots.
I use AI tools, including Claude, to help research and write. This story doesn't endorse any candidate or party.
Sources:
- The Washington Post: AI ad mocks Talarico. Democrats bet it will backfire.
- Yahoo News: Texas Senate race AI ads (Paxton campaign's defense)
- Yahoo News: Republicans lead AI ad spending (NYU research)
- Yahoo News: Fake scenes, real politics (Texas law and federal races)
- CNN via AOL: Republicans release AI deepfake of James Talarico
- Newsweek: Paxton AI attack ad
- Religion News Service: the 2021 "nonbinary" remark
- PolitiFact: Talarico AI ads fact-check
- Action Network: Texas Senate race final week (polling)
- Fox 4: complaint against Lt. Gov. Dan Patrick over AI ads
- Texas Tribune: Goodwin complaint and Patrick campaign response
- TribTalk: Texas deepfake law and the First Amendment
More to read
- From MKUltra to AI: the long history of messing with your mind
- Forced to pick, two big AIs split along party lines
- The first AI election: $137 million in AI-made ads, and the AI money behind the races
- How to spot an AI political ad before you vote (or share it)
- What Trump is actually doing about AI, and what Americans think of it
- Is AI raising your electric bill? Here's the honest answer.