Washington — As the 2024 election, the first since generative artificial intelligence became widely available, approached, experts feared the worst: Social media would be flooded with hyper-realistic AI-generated deepfakes, leaving confused voters unsure of what to believe.
So far, that hasn’t happened. Instead, voters are seeing something far more absurd: a video of former President Donald Trump holding an assault rifle while riding a cat; a mustachioed Vice President Kamala Harris dressed in Communist garb; and a passionate embrace between Trump and Harris.
AI plays a major role in presidential elections, but the biggest fears about it threatening the US presidential election have yet to come to fruition. Fake AI-generated images are circulating frequently on the web, but many of them are too cartoonish and absurd for even the most casual viewer to take them seriously.
But these memes also have a problem: Eye-catching AI-generated photos and videos, some of which are intended to be funny, have become useful tools for spreading false and sometimes racist messages that are clearly politically motivated — and are shared on social media by candidates and their supporters.
For example, Trump and many of his allies not only repeatedly promoted the unfounded conspiracy theory that Haitian immigrants were stealing and eating cats and dogs in Springfield, Ohio, but also spread related AI-generated memes. One meme shared by Trump’s Truth Social account showed Trump aboard a luxury jet surrounded by cats and white ducks. Another showed a group of kittens holding up signs that read, “Don’t let them eat us! Vote for Trump!”
Online propaganda expert Francesca Tripodi said such AI-generated images were a new viral tool for spreading old anti-immigrant messages.
“The memes that are amplifying this argument are not funny, and when we have public officials using this imagery as a means to perpetuate racism and xenophobia, that’s a big problem,” said Tripodi, a sociologist at the University of North Carolina at Chapel Hill.
Republicans have defended the images as light-hearted jokes and a by-product of Trump’s personality.
“The personality culture around Donald Trump encourages an exaggerated style of communication that turns things into comical memes,” said Republican strategist Caleb Smith. “The goal is to entertain, not to deceive, and that’s how it should be.”
Trump and his supporters are not the only ones creating AI memes, but they appear to be using AI image-generating tools more than Democrats. Some left-leaning users have posted AI images mocking billionaire Elon Musk, the owner of X and an ardent supporter of the Trump campaign. Democrats also posted AI-generated images of Trump being handcuffed and chased by police during his Manhattan court appearance last year.
However, Kamala Harris’ campaign has not leaned into spreading AI-generated content, instead sticking to TikTok trends and other memes that don’t require the creation of AI models.
“Currently, the only uses of generative AI permitted by the campaign are productivity tools like data analysis and industry-standard coding assistants,” said Mia Ehrenberg, a spokesperson for the Harris campaign.
Trump campaign spokesman Steven Chang did not respond to specific questions from The Associated Press but said the strategy had not changed since May, when the campaign issued an emailed statement saying it was “not using or engaging with” tools provided by any AI companies.
The use of fake, entertaining, and often absurd imagery to score political points is nothing new, but unlike a jumble of Photoshopped images or political cartoons, AI-generated images can have a stronger impact through their hyperrealism and bring new attention to political messages.
While some of the images of Springfield’s pets were cartoonish and humorous, many felt they promoted harmful conspiracy theories about the area, which subsequently received bomb threats and forced the evacuation of schools and government buildings.
“Memes that are clearly parody are one thing, but memes that are clearly intended to deceive are another,” said Rep. Adam Schiff, a California Democrat and vocal critic of Trump. “We’re already seeing the Trump campaign blur that line.”
The speed and accessibility of generative AI tools makes it easy to create quirky political content that drives clicks and likes. Accessible to anyone with an internet connection, AI image generators are a cheap and convenient way for campaigns to keep up with online trends and make their message stick.
“Campaigns have had to deal with disinformation and misinformation for a long time. … It’s not a new problem, but clearly AI allows them to do this more quickly, perhaps more persuasively and in a more targeted environment,” said Teddy Goff, who served as digital director for Barack Obama’s 2012 reelection campaign.
Paul Ingrassia, a New York-based political commentator and lawyer, said that at Glock’s urging, he created an image of Trump emerging from a lion’s den in a matter of seconds, then published it in his newsletter and sent it to Trump campaign staff. Trump’s Truth social account posted Ingrassia’s newsletter containing the image that day.
“My liaison to the president reached out to me and said, ‘The president loves this image. How did you make it? Who made it?’ And I said, ‘Yeah, I did. I made it for this story,'” Ingrassia says. “And the president said, ‘Keep up the great work. The president loves it.'”
The use of AI for political satire and propaganda is not limited to the US, but has been observed in elections from Indonesia to the Netherlands.
More sinister deepfakes have also attempted to influence elections around the world. Last year in Slovakia, an AI audio clip was seen impersonating the leader of the Liberal Party and talking about voter fraud just days before a congressional election. In January’s New Hampshire primary, an audio deepfake of President Joe Biden was sent in a robocall to Democratic voters, urging them not to vote. The incident was quickly made public and led to criminal charges.
Trump’s embrace of AI-generated imagery contradicts some of his past statements: In an interview with Fox Business earlier this year, he said AI is “very dangerous” and “very scary” because “there are no real solutions to the problems that are created by advancing technology.”
Some Republicans also worry that President Trump and the GOP are using AI to create political memes.
“I’m not into memes. I never have been into them and I never will be into them,” said Rep. Brian Fitzpatrick, a Republican who represents a battleground district outside Philadelphia. “I don’t believe in memes.”
___
Swenson reported from New York.
___
This story is part of “The AI Campaign,” an Associated Press series exploring the impact of artificial intelligence in the 2024 election cycle.
___
The Associated Press receives support from several private foundations to strengthen its commentary coverage of elections and democracy. Learn more about the AP Democracy Initiative here. The Associated Press is solely responsible for all content.
The Associated Press receives financial support from the Omidyar Network to support coverage of artificial intelligence and its impact on society. The AP is solely responsible for all content. For AP’s charity engagement criteria, a list of supporters and areas of coverage funded, visit AP.org.
