TThe New York mayoral election may be remembered for the young democratic socialists’ remarkable victory, but it was also marked by something that is likely to permeate future elections: the use of AI-generated campaign videos.
Andrew Cuomo, who lost last week’s election to Zoran Mamdani, has shown particular interest in sharing deepfake videos of his opponent, including one in which the former governor is accused of racism, in a developing area of campaigning.
AI has been used in election campaigns before, particularly when using algorithms to target specific voters and, in some cases, to develop policy proposals. But as AI software develops, it is increasingly being used to create sometimes misleading photos and videos.
“I think what was really groundbreaking this election cycle was the use of generative AI to generate content that goes directly to voters,” said New York State Representative Alex Boas, who has been at the forefront of introducing legislation regulating the use of AI.
“So whether it was the Cuomo campaign using ChatGPT to create a housing plan, or Mr. Cuomo and a number of other campaigns creating AI-generated video ads for voters, I mean, I think it felt very new in the 2025 cycle, or certainly felt far ahead of anything we’ve seen before.”
Incumbent Mayor Eric Adams, who withdrew from the race in September, used AI to create robocalls to New Yorkers speaking Mandarin, Urdu, and Yiddish, and also produced an AI video depicting New York as an apparent war-torn dystopia to attack Mandani.
Meanwhile, Mr. Cuomo was accused of racism and Islamophobia after his campaign tweeted a video showing the fictional Mr. Mamdani eating rice with his fingers and a black man shoplifting. The ad also features a black man wearing a purple shirt and tie, a fur coat, and holding a silver cane, appearing to endorse sex trafficking. Cuomo’s campaign later deleted the document, claiming it had been sent in error.
Boas, who is running to represent New York in the House of Representatives, said many of the AI-generated ads in the last election cycle were “likely to veer into what we would consider to be bigoted territory.”
“I think this is something we need to track as well: Is this because the algorithm is reproducing the stereotypes that are in the training data? [is it] Because it is very easy to operate. You don’t have to tell an actor of a certain race to do a certain thing, you just change it in the computer,” Boas said.
“You don’t have to tell someone to their face to express themselves a certain way. It actually makes it easier for people to put out content that they think polite society should frown upon.”
In New York state, campaigns are supposed to label AI ads as such, but some ads, including one posted and removed by Mr. Cuomo, were not labeled as such. The New York State Board of Elections is in charge of potential charges against campaigns, but Boas noted that campaigns may be willing to respond to any penalties, especially if they are imposed after the campaign is over.
“I think you’re going to find a camp that’s willing to take that tradeoff. If they win, they don’t care if they pay the fine afterwards, and if they lose, it doesn’t matter,” Boaz said. “So we want to find an enforcement system that can quickly de-escalate the situation before the election, rather than just punishing them after the election.”
Robert Wiseman, co-director of Public Citizen, a nonprofit advocacy group that has been involved in passing many AI laws across the country, said trying to deceive people is now illegal in more than half of states, and campaigns are required to include a disclaimer on generated AI ads that they are not genuine. Still, he said regulating the use of AI in campaigns is a pressing issue.
“Lying has been a part of politics since time immemorial. This is different from lying, and it’s different from saying you said something you didn’t say,” Wiseman said.
“When you’re shown an obviously authentic version of someone saying something, it’s very difficult for that person to counter that and say, ‘I didn’t say that,’ because you’re asking people not to believe what they see with their own eyes.”
AI is now able to generate reliable videos, but some campaigns aren’t quite able to do so. The “Zoran Halloween Special” video posted by Cuomo (which the ad explicitly said was generated by AI) showed a very sloppy production of Mamdani, complete with out-of-sync audio and an incomprehensible script.
With the midterm elections approaching and the 2028 presidential election looming, AI-generated political videos are likely to be here to stay.
They are already used at the national level. Elon Musk shared an AI-generated video in July 2024 after Kamala Harris became the de facto presidential candidate. The video showed Harris claiming to be the “ultimate diversity recruit” and saying she “doesn’t know the first thing about running a country.”
While states may be making strides in regulating the use of AI in elections, there appears to be little appetite to do so at the federal level.
During the No Kings protests in October, Donald Trump shared an AI video that showed him piloting a fighter jet and dropping brown liquid on Americans, but this is just the latest in a series of AI video posts by him.
President Trump clearly approves of this medium, so it seems unlikely that Republicans will try to rein in AI any time soon.
