NRSC airs first fully AI-recreated candidate deepfake in US Senate race ad

The National Republican Senatorial Committee released an 85-second attack advertisement built around a computer-generated recreation of Texas Democratic Senate candidate James Talarico, reading aloud and commenting on his own past social-media posts from 2013 and 2021 touching on transgender rights, race, religion and Planned Parenthood. Unlike earlier political ads that used AI only for backgrounds or brief effects, the clip presented an artificial version of the candidate himself for its full duration, which several outlets reported as the first ad of its kind in a US Senate race to do so.

The video carried an “AI GENERATED” disclosure, but as small text confined to a corner of the screen for most of its length — a labelling approach that digital-forensics researcher Hany Farid, cited in coverage of the ad, questioned as inadequate for viewers scrolling quickly through social media. The NRSC defended the ad as a straightforward visualisation of Talarico’s own authentic statements using new technology rather than as fabricated speech. Talarico’s campaign called the ad an attempt to mislead voters, with a spokesperson saying opponents were “scared of James Talarico.”

The ad drew attention less because AI-generated political content was new by 2026 than because of its scale and format: a full-length synthetic recreation of an opposing candidate’s face and voice, deployed by a major national party committee rather than an outside group, ahead of the 2026 midterms. It added to renewed calls from critics, including consumer advocacy groups, for federal rules on AI use and disclosure in campaign advertising, an area where the US had no comprehensive federal statute as of early 2026.