San Francisco, USA: Bruce Blakeman, a Republican running for governor of New York, released campaign ads featuring New York City Mayor Zohran Mamdani and state Governor Kathy Hochul gardening together and collaborating to raise prices for local residents. Another ad showed Hochul discussing steep increases in tolls and transit fares. Neither official participated in the videos, which were entirely generated using artificial intelligence.
A wave of obviously fake AI content has become a prominent feature of the lead-up to the pivotal United States midterm elections on November 3. Although much of the regulatory discussion centers on misleading deepfakes, experts note that the current election cycle is heavily defined by identifiably fake material known as "slopaganda."
Data from the non-partisan Wesleyan Media Project shows that political candidates and their supporters have spent more than 80 million dollars to produce over 164 AI-based advertisements this season. Republican candidates and their backers account for roughly 80 percent of these materials.
Blakeman defended his campaign strategy by asserting that the videos carry brief initial disclaimers identifying them as AI-generated and align with a long tradition of American political satire. However, researchers point out that this mixture of low-quality AI content and propaganda is designed to communicate deeper emotional narratives rather than factual accuracy, potentially impacting voter trust and candidate perception.
Other notable examples include advertisements depicting Ohio Democratic Senate candidate Sherrod Brown as an unwelcome party guest, Texas Democratic Senate candidate James Talarico wearing a dress reminiscent of The Sound of Music, and Michigan Republican Senate candidate Mike Rogers portrayed as a superhero.
In San Francisco, Democratic state Senator Scott Wiener faced immediate backlash after releasing an AI chatbot during his high-profile campaign to succeed retiring House Speaker Nancy Pelosi. The interactive tool answered thousands of inquiries regarding his opponent, Connie Chan, while parodying her record on real estate development, accent, and naturalized citizenship. The chatbot drew criticism for perceived bigotry, leading Wiener to withdraw the tool and acknowledge that it missed the mark.
The proliferation of these tools occurs within a legal framework designed to safeguard political speech and parody. While federal legislation does not explicitly regulate AI deepfakes in campaigns, 31 states have enacted laws requiring artificial intelligence-generated advertisements to feature clear disclosures.
Advocacy groups continue to advocate for comprehensive federal standards that balance free expression with voter transparency, emphasizing that labels must become as instantly recognizable as nutritional information. As the election approaches, campaigns continue to grapple with how emerging technologies are reshaping political discourse and voter evaluation.