AI Deepfakes Invade US Midterms: Fake Ads Flood the Information Space

San Francisco, US: Bruce Blakeman, a Republican candidate running for governor of New York, has released campaign ads featuring Zohran Mamdani, the mayor of New York City, and Kathy Hochul, the state governor, gardening together and working in tandem to raise prices for residents. Another ad shows Hochul talking about raising bridge tolls and subway fares so that people cannot afford to drive or take the train.

In reality, neither Hochul nor Mamdani appeared in these ads, which were created with artificial intelligence. A wave of AI-generated ads, plainly fake yet capturing voters' attention, has become a hallmark of campaigns ahead of the pivotal US midterm elections on November 3.

According to Scott Babwah Brennen, director of the Center for Technology Policy at New York University and co-author of a recent report titled "Political Slopaganda," while regulations focused on labeling AI deepfakes may confuse voters, it is "slopaganda" itself — recognizable fake advertising — that defines this election season.

According to the Wesleyan Media Project, a nonpartisan election ad tracker, political candidates and their supporters have spent more than $80 million to create over 164 AI-based ads this election season. The group found that Republican candidates and their supporters were behind 80 percent of those ads.

Blakeman defended his ads, saying they are clearly labeled as AI-generated — a disclaimer the ads display for a few seconds — and follow a long tradition of satire and parody in American politics.

However, slopaganda — a blend of low-quality AI "slop" and propaganda — is being used to "communicate deeper truths rather than objective truths," according to Brennen. That means it can shape perceptions of a candidate's character or ideology and can "affect voter trust."

Other similar ads show Sherrod Brown, a Democratic Senate candidate from Ohio, as the last guest at a party who has stayed too long. James Talarico, a Democratic Senate candidate from Texas, is depicted in a blue dress resembling the character Maria in The Sound of Music, while Mike Rogers, a Republican Senate candidate from Michigan, appears as a superhero.

In San Francisco, when Democratic California state Senator Scott Wiener launched an AI chatbot in the high-profile race to succeed honorary House Speaker Nancy Pelosi, who represents the tech city in Congress, it immediately sparked a fierce backlash. The chatbot answered roughly 5,000 questions about his opponent, Connie Chan, mimicking her record of allegedly blocking real estate development in the city, as well as her voice and naturalized citizenship. The move drew criticism as discriminatory and Wiener withdrew the chatbot, saying that although it was innovative, "it did not achieve its goal."

No federal law

The deepfake ads have emerged against a backdrop of protections for political free speech, including when expressed through parody. Although there is no federal law governing ads created with AI deepfakes, 31 states have passed their own laws requiring labeling for ads that use AI.

"State laws do what they can without infringing on the First Amendment," said Ilana Beller, director of state AI governance and policy at Public Citizen, a consumer advocacy group based in Washington, DC. "Federal law is still needed," she said.

Adam Rose, executive director of Starling Lab, a research center on technology and human rights co-founded by Stanford University and the University of Southern California, argues that some AI-driven changes may be nothing more than color or voice edits that governments need not require labeling for. He said AI-generated ads should not be banned.

Instead, Rose said, candidates have always made "judgment calls" about advertising, and those decisions reflect on them. For Wiener, Rose said, releasing the chatbot "shows poor judgment. It shows voters something about him."

Wiener is running to represent San Francisco in the US Congress, a city that is a global epicenter of the AI industry, with two of the largest AI companies, Open AI and Anthropic, headquartered there. The city's representative will play an important role as AI legislation emerges, especially as polls show a majority of Americans are more concerned than excited about the technology.

"This is a transitional moment when AI is reshaping our society," said Ian Krager, a spokesman for Connie Chan. "This race is a harbinger for the whole country. This is the moment we can show the world that America can lead in AI development without asking workers to bear the cost."

"And it's clear Connie Chan stands with working families while Scott Wiener stands with AI companies," he said, referring to the AI chatbot.

As a state senator, Wiener drafted a California law last year requiring AI companies to disclose security incidents involving autonomous systems created by extremely large AI models that engage in dangerous or harmful incidents. A less stringent version became law last year. His office did not respond to Al Jazeera's request for comment.

Labels as clear as food labels

Looking toward the final stretch of the November midterms and beyond, Beller of Public Citizen said labeling of ads that use AI needs to be as clear and easy to understand as food labels so voters can assess them properly. "People need to know where to look for the label," she said, because it is important to build awareness of AI deepfake ads, especially among young voters and older people.

In New York, Democrats filed a complaint with the state election board saying the ads violate disclosure requirements and "confuse voters." But Blakeman has said that "people find it funny. They like satire."

The ads are "trying to grab attention when candidates know there is political power to be gained by grabbing attention," said Brennen of NYU.

With AI deepfake ads on the rise, some candidates have publicly said they will not use them. "People seem to like [candidates] because of that," Beller said.