AI deepfake ads proliferate in US midterm elections: federal regulation remains absent, Republican camp accounts for 80%

[Press report] According to Al Jazeera, statistics from the nonpartisan election advertising tracker Wesleyan Media Project show that during the current 2026 US midterm election season, candidates and their supporters have spent more than $80 million to produce at least 164 political advertisements based on artificial intelligence. The Wesleyan Media Project found that 80% of that content came from Republican candidates and their supporters. This share, layered on top of the long-standing absence of federal legislation regulating AI political advertising in the United States, has deepened outside concerns about the authenticity of election information and the effectiveness of regulation.

New York State Republican gubernatorial candidate Bruce Blakeman was the first to push this new campaign tool to the center of public attention. According to Al Jazeera, an advertisement released by Blakeman featured footage showing New York Governor Hochul and New York City Mayor Mamdani working together in a garden, with dubbed audio in which they discussed raising prices, raising tolls and car fees. Neither Hochul nor Mamdani actually took part in the filming of the advertisement, and the relevant footage was entirely generated by AI. Blakeman argued that the ad was clearly labeled as "AI-generated" at its opening, consistent with the American tradition of political satire and parody.

Scott Babwah Brennen, director of the Center for Technology Policy at New York University, told Al Jazeera that this kind of content — which he termed "slopaganda," a hybrid of low-quality AI "slop" and political propaganda — is aimed not at conveying factual truth but at conveying a "deeper truth," and may sway voters' perceptions of a candidate's character and ideology, ultimately eroding voter trust in elections.

On the regulatory front, according to Al Jazeera's review, the United States currently has no federal law governing AI deepfake advertisements. Thirty-one states have passed local laws requiring labels on AI-generated advertisements, but the standards vary from state to state and the coverage is limited. Ilana Beller, the state AI governance and policy lead at Public Citizen, a Washington, D.C.-based consumer advocacy organization, noted: "State-level laws are doing the best they can without running afoul of the First Amendment," but she also stressed that "federal-level legislation is still needed."

The Democratic camp has likewise been drawn into controversy. California Democratic state Senator Scott Wiener, in a key race to succeed retiring House Speaker Pelosi in Congress, rolled out an AI chatbot targeting his opponent Chen Yu-chen. The chatbot responded to roughly 5,000 questions in a mocking tone, ridiculing Chen's record of blocking housing development in San Francisco, her accent and her naturalized citizenship. The tool sparked backlash over alleged bias, and Wiener subsequently withdrew it, acknowledging that while innovative, "it did not meet the goal." Adam Rose, chief operating officer of the Sterling Lab jointly founded by Stanford University and the University of Southern California, said the move "displayed poor judgment and also showed voters something about the candidate himself." Wiener himself had led the drafting last year of a California law requiring AI companies to disclose major safety incidents involving autonomous systems; a watered-down version of the bill ultimately passed.

The New York State Democratic Party has filed a complaint with the state elections board, alleging that Blakeman's AI ad violated disclosure requirements and was "misleading to voters." Blakeman responded that "people find it funny, they like satire." The state elections board's ruling on the complaint has not been publicly disclosed.

Al Jazeera's report also catalogued other typical cases: Ohio Democratic Senate candidate Sherrod Brown was depicted as the last guest who would not leave a party; Texas Democratic Senate candidate James Talarico was placed in a scene featuring a blue dress reminiscent of the lead character Maria in "The Sound of Music"; Michigan Republican Senate candidate Mike Rogers appeared in a superhero image. In an interview, Beller said that as AI deepfake advertisements continue to grow, labeling rules need to be as clear and easy to understand as food labels, so that voters can make judgments based on them. She also noted that some candidates have publicly pledged not to use such content, and that "people seem to support them more because of it."

Citing analysts, Al Jazeera observed that, in the absence of a unified federal standard, the use or non-use of AI-generated ads is becoming a public test of a candidate's "judgment" — which is precisely the core reason such low-cost, highly viral content has been able to rapidly permeate American electoral discourse. With the November 3 midterm election approaching, it remains uncertain whether federal-level AI advertising regulation will be enacted before voting day.