A Senate Candidate Deepfaked in His Own Race. The Fake Ads Are Multiplying.
Artificial intelligence has become a leading issue in the U.S. midterm campaign for the first time, a Washington Post analysis found on Aug. 14, 2026 — and the same tools are fabricating the candidates themselves. A lifelike AI fake of Texas Senate hopeful James Talarico is one of at least 15 AI-generated ads to reach the 2026 races, even as federal rules to police them stall.
With the November midterm elections approaching, artificial intelligence has moved to the center of American politics. A mid-August analysis by The Washington Post found that more candidates now discuss AI or data centers on their campaign websites than mention long-standing issues such as Israel or manufacturing — the first time the technology has registered as a top-tier campaign concern.
The money has followed. Competing super political action committees backed by the AI industry are already pouring funds into the 2026 midterms, beginning with the year’s first primaries, according to The Washington Post. But the sharper worry for election officials is not what the technology is spending on — it is what it can now fabricate.
A candidate, faked
The clearest example came from the National Republican Senatorial Committee, which released an ad depicting an AI-generated version of James Talarico, the Democratic nominee for U.S. Senate in Texas, speaking directly to camera for more than a minute as he read aloud his own years-old social-media posts on race, transgender rights and religion, CNN reported. It was, the network noted, the first such ad to feature a phony version of a candidate talking in a lifelike manner for so long — a marker of how quickly the tools have advanced.
The video carried a disclosure — the words “AI generated” — but in small, easy-to-miss type confined to a corner of the screen, according to CNN. It was not an isolated case. At least 15 campaign ads featuring AI-generated content have run since November, NBC News reported, with the technology appearing in contests from school-board campaigns to governor’s races, including a Massachusetts case in which an ad mimicked a rival politician’s voice.
The label voters miss
Researchers who study political communication say the danger lies in the gap between how convincing the fabrications have become and how faint the warnings attached to them are. Where disclosures exist at all, they are often small enough to slip past a scrolling viewer, NBC News reported. The result, critics warn, is content designed to be shared and believed before it can be checked — the same dynamic that has driven online disinformation on social platforms for years, now applied to the words and faces of candidates themselves. The synthetic clips travel on the same engagement-ranked feeds, where a fabricated video can circulate widely before any correction catches up.
A patchwork of rules
Regulation has not kept pace. About half of U.S. states — 26 — have laws governing political deepfakes, either requiring disclosure or barring their use in a window before an election, NBC News reported. The Washington Post’s August analysis put the figure at roughly half the states, a review of campaign data found, leaving a checkerboard of protections that vary sharply from one race to the next.
At the federal level, there is no law requiring AI disclosures on political advertising, and bills to create one have repeatedly failed to reach a vote, NBC News reported. The Trump administration, meanwhile, has moved to curb state AI regulation more broadly, signing an executive order pressuring states not to regulate the technology and directing officials to challenge existing state laws, CBS News reported. The order exempts some measures, such as child-safety protections, but its central aim is a lighter-touch national framework that would override state rules, PBS News reported — leaving the fastest-moving part of the campaign, for now, the least governed.