With the US midterm elections approaching on November 3, the use of fake political advertisements created using artificial intelligence (AI) is on the rise. These ads depict real politicians saying things or performing actions they never actually did. While it is often obvious that the ads are AI-generated, they still succeed in capturing voters’ attention and shaping perceptions of the candidates’ images and political stances.
In a campaign ad for Republican New York gubernatorial candidate Bruce Blakeman, New York City Mayor Zohran Mamdani and State Governor Kathy Hochul are shown gardening together. The ad portrays them as agreeing on a plan to increase costs for residents.
Another ad shows Hochul stating that she wants to raise tolls and transit fares to the point where people can no longer afford to drive or use the subway.
However, neither Hochul nor Mamdani participated in these advertisements; they were created using artificial intelligence.
According to Scott Babwa Brennen, Director of the Center for Technology and Society at New York University, this election cycle has seen a prevalence of ads that are clearly fake upon inspection. He has termed this type of campaigning “slopaganda”—a portmanteau combining “slop” (referring to low-quality AI-generated content) and “propaganda.” Brennen is a co-author of the recently published report titled “Political Slopaganda.”
Data from the Wesleyan Media Project indicates that political candidates and their supporters have spent over $80 million creating more than 164 AI-generated ads this election season. The non-partisan organization, which monitors election advertising, reports that Republican candidates and their supporters are behind 80 percent of these ads.
Blakeman has defended his advertisements. He maintains that the advertisements clearly state they were created using AI. A warning to this effect also appears for a few seconds at the beginning of the ads. In his view, US politics has a long tradition of satire and parody, and these advertisements are part of that lineage.
However, Brennan believes that the goal of such campaigns is not always to spread outright falsehoods. Instead, they aim to shape perceptions regarding a candidate’s character, ideology, or political stance. As he puts it, these ads claim to reveal a “truth deeper than factual truth.” Yet, they can influence voter trust and political assessments.
**Some as unwelcome guests, others as superheroes**
AI-generated satirical ads depict various candidates in unusual roles. In the Ohio Senate race, Democratic candidate Sherrod Brown is portrayed as a party guest who has overstayed his welcome and whom no one wants around anymore. In Texas, Democratic candidate James Talarico appears in a blue outfit reminiscent of Maria from the film *The Sound of Music*, while in Michigan, Republican Senate candidate Mike Rogers is depicted as a superhero.
A controversy also arose in San Francisco, California, over an AI-generated chatbot. Democratic State Senator Scott Wiener, who was running for the US Congress, launched the chatbot to answer questions about his rival, Connie Chan.
The chatbot answered approximately 5,000 questions. It used satire to accuse Chan of obstructing housing and infrastructure projects in the city and mocked her accent and citizenship status. These actions drew accusations of bias and hateful conduct. Facing criticism, Wiener withdrew the chatbot, acknowledging that while the initiative was innovative, it failed to meet expectations. Adam Rose, Chief Operating Officer of the Starling Lab—a technology and human rights research center established jointly by Stanford University and the University of Southern California—states that AI-generated advertisements should not be banned outright. This is because some AI modifications might involve only minor adjustments, such as color correction or voice tweaking, which may not warrant a mandatory government warning.
However, according to Rose, the decision to use AI in advertisements rests with the candidates themselves. Such decisions reflect their judgment and shape voters’ perceptions of them. Citing the example of Weiner’s chatbot, he noted that the initiative conveyed a message to voters regarding Weiner’s decision-making capabilities.
**No Federal Law on AI Advertisements**
There is currently no federal law in the United States regulating AI-generated political advertisements. However, 31 states have enacted their own laws requiring the disclosure of AI usage in such advertisements.
Ilana Beller, Director of the AI Policy Division at Public Citizen—a Washington-based consumer rights organization—notes that states are drafting these laws while keeping the First Amendment’s protection of free speech in mind.
Nevertheless, she believes a federal law addressing AI-generated deceptive advertisements is necessary.
Adam Rose of the Starling Lab also holds the view that, rather than imposing a blanket ban on AI-generated advertisements, the manner in which they are utilized should be taken into account. Meanwhile, Beller emphasizes the need to ensure that voters can easily recognize when AI is being used in advertisements.
Ian Kreger, a spokesperson for Connie Chan—Wiener’s rival in the race for the San Francisco congressional seat—stated that this election could set a significant precedent for the United States.
He believes the U.S. can lead the world in AI development, provided the costs are not borne by working-class people. He alleges that Chan stands with working-class families, whereas Wiener aligns with AI companies.
Last year, Wiener spearheaded an initiative in California to enact legislation requiring AI companies to disclose safety incidents resulting from automated systems powered by massive AI models. A somewhat diluted version of the bill was eventually signed into law.
Wiener’s office did not respond to Al Jazeera’s request for comment on this matter.
**Call for Clear Labels, Like Food Packaging**
Looking ahead to the final stages of the November midterm elections and beyond, Ilana Beller of Public Citizen advocates for clearly labeling political advertisements that utilize AI.
She argues that just as food packaging clearly lists ingredients, ad labels should be transparent and easy to understand, enabling voters to verify information and make informed decisions.
Beller emphasizes the urgent need to raise awareness about AI-generated fake ads, particularly among young and elderly voters, noting that people must be taught to look for these labels.
In New York, the Democratic Party has filed a complaint with the state’s Board of Elections regarding advertisements by Blackman. The party claims the ads violate disclosure rules and mislead voters. Blackman, however, maintains that people view the ads as humorous and appreciate the satire.
Brennan notes that candidates recognize the political advantage of capturing public attention, which is driving the rise of AI-generated ads. Conversely, some candidates have publicly declared that they will not use such advertisements. According to Belar, people are also showing a positive attitude towards candidates who have decided not to use AI-generated advertisements.
