A.I. Attack Ads Flood U.S. Campaigns, From North Carolina Yoga Studios to Statewide Races
A surge of AI‑generated political ads is reshaping U.S. elections, with dozens of costly spots targeting candidates across the country.
AI‑Powered Attack Ads Hit a North Carolina Senate Race
In August, the re‑election campaign of North Carolina state senator Michael Lee released a television spot titled “Bonkers Bichler.” The ad attacks his Democratic opponent, yoga‑studio owner Jessica Bichler, by showing a hyperrealistic, animated AI rendering of her doing yoga. The figure on screen is not Bichler herself; the AI version twists into an impossible knot while smiling vacantly.
Bichler said she was “thinking, who was even advising on this? Because it was really bizarre” when she first saw the footage. A disclaimer appears five seconds in, noting that AI was used to make “these silly video clips,” yet some local voters still asked her why she had let Lee into her studio.
She warned that AI allows people to make any image of you doing anything they want … and put that out in the public sphere, calling the technology a “huge vulnerability.” The ad also carried a hint of hypocrisy, as Lee and his wife are known to frequent yoga studios themselves.
AI Imagery Spreads to Other North Carolina Races
Bichler has observed similar AI‑generated images in ads targeting other state Senate candidates, including April Cook and James Gailliard. Lee himself has been impersonated by a group unrelated to his campaign, which placed cardboard cutouts of him holding money and wearing a shirt that reads I <3 special interest, with a caption suggesting he “peed his pants.”
National Scale of AI Political Advertising
AdImpact, a political‑advertising tracker, has identified fifty‑eight unique AI ads across the United States, with total spending exceeding ten million dollars. Lee’s campaign alone spent more than one hundred thousand dollars on the “Bonkers Bichler” spot and two follow‑up ads.
Recent History of AI Deception in Elections
AI‑generated political content is not new. In January 2024, thousands of New Hampshire voters received a recorded message that sounded like President Joe Biden urging them not to vote in a primary. The deepfake was created with ElevenLabs by magician Paul Carpenter for a consultant linked to Dean Phillips’s presidential campaign.
In October 2025, Andrew Cuomo’s mayoral campaign posted AI spots targeting Zohran Mamdani, including a caricature as an “Austin Powers” Mini‑Me and a silent figure eating rice with his hands. The videos were criticized as racist, removed quickly, yet continued to circulate online.
AI Ads Compared to Traditional Attack Tactics
A recent AI‑generated ad against Pennsylvania governor Josh Shapiro (by opponent Stacy Garrity’s campaign) shows a realistic AI version of Shapiro speaking about “deleting incriminating e‑mails” and “supporting boys in girls sports.” Unlike older ads that used actors playing ordinary voters, the AI version places the opponent’s own likeness into the message.
Manuel Bonder, spokesperson for Shapiro, called the spot “cheap, dishonest AI bullshit” and highlighted Garrity’s stance on “aggressive data center deregulation.”
Republican Dominance and Thematic Targets
Two‑thirds of the AI ads identified by AdImpact come from Republican campaigns, often focusing on personal identity. Examples include:
A Michigan Senate ad depicting candidate Abdul El‑Sayed playing a board game with his sister and streamer Hasan Piker, with cards flashing “DEATH TO AMERICA.”
A Texas Senate ad showing Democrat James Talarico in a T‑shirt supporting trans people and eating tofu at a football game, implying a vegan lifestyle.
A Georgia Senate ad portraying Jon Ossoff in a salon, with technicians filing his hands rather than his nails, suggesting hidden motives.
Regulatory Landscape and Legal Responses
Current law treats AI political ads as protected speech, similar to political cartoons. The 2025 Take It Down Act forces platforms to remove sexually explicit deepfakes but does not cover political AI content.
After the 2024 fake‑Biden robocall, consultant Steve Kramer was removed from the campaign, faced criminal charges for voter suppression and impersonation, and was fined by the FCC, though he was later acquitted. The FCC subsequently banned AI‑generated voices in robocalls.
Bichler’s campaign sent a cease‑and‑desist letter to Lee’s campaign, alleging misrepresentation of her nonprofit affiliation. She noted that “there are no laws on the books to say he can’t do that,” referencing Lee’s vote to table a disclosure amendment for generative AI use in media.
Impact on Voters and the Political Conversation
Bichler worries that even voters who recognize AI fakes may be subtly swayed by the imagery. She said voters already live in algorithmic bubbles, and AI attack ads “provide further fuel for that fragmentation,” leaving people talking with conviction because “I’ve seen this with my own eyes.”
The flood of AI‑generated political ads marks a tipping point for campaign tactics, raising questions about truth, regulation, and the future of democratic discourse.
