ERIC WESTERVELT, HOST:
AI-generated images have been a thing this election season. Deep fakes haven't necessarily been fooling people, at least not for long, but experts watching online spaces say the images serve a different purpose - propaganda. NPR's Huo Jingnan has our story.
HUO JINGNAN, BYLINE: You might have seen this image circulating on social media already.
EDDIE PEREZ: This AI-generated image of a very cute and very helpless-looking little girl holding a puppy in a canoe in the midst of what appears to be terrible flooding associated with the recent hurricanes.
JINGNAN: That was Eddie Perez, a former Twitter employee who is now focusing on confidence in elections at the non-partisan nonprofit OSET Institute. While the image got millions of views on X, formerly Twitter, many users were quick to point out that it carries telltale signs of generative AI. But that didn't matter for some prominent people who shared it like Amy Kremer, who is a Republican national committee member representing Georgia. She wrote that she, quote, "doesn't care where the image came from." It's, quote, "emblematic of real suffering."
RENEE DIRESTA: As a form of political communication, you can use an image to convey a broader truth.
JINGNAN: Renee DiResta is a researcher at Georgetown University who recently published a book about online influencers. She says that the so-called deeper truth can be divorced from facts.
DIRESTA: It's a form of political propaganda, a way to signal interest and support for a candidate, almost like in a fandom kind of style.
JINGNAN: Many Trump supporters expressed sympathy for the fictional girl, but it was also a way to criticize the Biden administration's hurricane relief efforts. A similar dynamic appeared after Trump and his running mate, JD Vance, boosted false claims about Haitian immigrants in Springfield, Ohio. Trump fans posted AI-generated images of the former president cuddling cats and ducks while attacking Biden's immigration policies.
DIRESTA: Political campaigns then can pick it up, can retweet it, can boost it, and are then seen as being sort of in on the conversation, maybe in on the joke themselves.
JINGNAN: This kind of participation gives supporters camaraderie and entertainment. It's not a new dynamic, but AI tools make it easy and potentially lucrative.
DIRESTA: On X, for example, you can monetize your account, and so if you produce a compelling image that a lot of people engage with, you actually can earn money from that.
JINGNAN: These images often don't stand up to scrutiny, but can pass as a photo at first glance. Studies show that people are bad at telling fakes from real photos. Take a recent image that falsely suggested popstar Taylor Swift had endorsed Trump. There was a lot of blowback after Trump shared it, even though it didn't really look like her.
ARA MERJIAN: It wouldn't have been a scandal if someone had drawn Taylor Swift in a comic endorsing Trump.
JINGNAN: Ara Merjian is an art history scholar at New York University. Swift has since endorsed Kamala Harris for president. In many ways, these AI-generated images aren't meant to pass as photos. They're more like political comics. Take this image.
PEREZ: A person who looked very much like Come Harris wearing a kind of Soviet-style or Red China-style, CCP, you know, communist-looking uniform.
JINGNAN: That's Eddie Perez again. He says these kind of images are still harmful, not because they are deceptive, but because they inflame political polarization.
PEREZ: These images and the polarization is sending the message that there is no legitimate way that Kamala Harris and her party could actually win a presidential election.
JINGNAN: And that could damage the public's confidence in the vote count. Huo Jingnan, NPR News. Transcript provided by NPR, Copyright NPR.
NPR transcripts are created on a rush deadline by an NPR contractor. This text may not be in its final form and may be updated or revised in the future. Accuracy and availability may vary. The authoritative record of NPR’s programming is the audio record.
