Get updates delivered to you daily. Free and customizable.
The Atlantic
The MAGA Aesthetic Is AI Slop
By Charlie Warzel,
2 days ago
Taylor Swift fans are not endorsing Donald Trump en masse. Kamala Harris did not give a speech at the Democratic National Convention to a sea of communists while standing in front of the hammer and sickle. Hillary Clinton was not recently seen walking around Chicago in a MAGA hat. But images of all these things exist.
In recent weeks, far-right corners of social media have been clogged with such depictions, created with generative-AI tools. You can spot them right away, as they bear the technology’s distinct image style : not-quite-but-almost photorealistic, frequently outrageous , not so dissimilar from a tabloid illustration . Donald Trump—or at least whoever controls his social-media accounts—posted the AI-generated photo of Harris with the hammer and sickle, as well as a series of fake images depicting Taylor Swift dressed as Uncle Sam and young women marching in Swifties for Trump shirts. (This after he falsely claimed that Harris had posted an image that had been “A.I.’d”—a tidy bit of projection.)
Trump himself has been the subject of generative-AI art and has shared depictions of himself going back to March 2023. He’s often dressed up as a gun-toting cowboy or in World War II fatigues, storming a beach. Yet these are anodyne compared with much of the material created and shared by far-right influencers and shitposters. There are plenty of mocking or degrading images of Harris and other female Democratic politicians, such as Alexandria Ocasio-Cortez. On X, one post that included a fake image in which Harris is implied to be a sex worker has been viewed more than 3.5 million times; on Facebook, that same post has been shared more than 87,000 times. One pro-Trump, Elon-Musk-fanboy account recently shared a suggestive image depicting a scantily clad Harris surrounded by multiple clones of Donald Trump; it’s been viewed 1.6 million times. There are images and videos of Harris and Trump holding hands on a beach and Harris wearing a crown that reads Inflation Queen . On the first night of the DNC, MAGA influencers such as Catturd2 and Jack Posobiec supplemented their rage tweets about Democrats with stylized AI images of Tim Walz and Joe Biden looking enraged.
Although no one ideology has a monopoly on AI art, the high-resolution, low-budget look of generative-AI images appears to be fusing with the meme-loving aesthetic of the MAGA movement. At least in the fever swamps of social media, AI art is becoming MAGA-coded. The GOP is becoming the party of AI slop.
AI slop isn’t, by nature, political. It is most prevalent on platforms such as Facebook, where click farmers and spammers create elaborate networks to flood pages and groups with cheap, fake images of starving children and Shrimp Jesus in the hopes of going viral, getting likes, and picking up “creator bonuses” for online engagement. Jason Koebler, a technology reporter who has spent the past year investigating Facebook’s AI-slop economy, has described the deluge of artificial imagery as part of a “zombie internet” and “the end of a shared reality,” where “a mix of bots, humans, and accounts that were once humans but aren’t anymore interact to form a disastrous website where there is little social connection at all.”
What’s going on across the MAGA internet isn’t exactly the same as Facebook’s spam situation, although the vibe is similar. MAGA influencers may be shitposting AI photos for fun, but they’re also engagement farming, especially on X, where premium subscribers can opt in to the platform’s revenue-sharing program. Right-wing influencers have been vocal about these bonuses, which are handed out based on how many times a creator’s content is seen in a given month. “Payout was huge. They’ve been getting bigger,” Catturd2 posted this March, while praising Musk.
Although many of these influencers already have sizable followings, AI-image generators offer an inveterate poster the thing they need most: cheap, fast, on-demand fodder for content. Rather than peck out a few sentences complaining about Biden’s age or ridiculing Harris’s economic policies, far-right posters can illustrate their attacks and garner more attention. And it’s only getting easier to do this: Last week, X incorporated the newest iteration of the generative-AI engine Grok, which operates with fewer guardrails than some competing models and has already conjured up untold illustrations of celebrities and politicians in compromising situations.
It’s helpful to think of these photos and illustrations not as nefarious deepfakes or even hyper-persuasive propaganda, but as digital chum —Shrimp Jesus on the campaign trail. For now, little (if any) of what’s being generated is convincing enough to fool voters, and most of it is being used to confirm the priors of true believers. Still, the glut of AI-created political imagery is a pollutant in a broader online information ecosystem. This AI slop doesn’t just exist in a vacuum of a particular social network: It leaves an ecological footprint of sorts on the web. The images are created, copied, shared, and embedded into websites; they are indexed into search engines. It’s possible that, later on, AI-art tools will train on these distorted depictions , creating warped, digitally inbred representations of historical figures. The very existence of so much quickly produced fake imagery adds a layer of unreality to the internet. You and I, like voters everywhere, must wade through this layer of junk, wearily separating out what’s patently fake, what’s real, and what exists in the murky middle.
In many ways, political slop is a logical end point for these image generators, which seem most useful for people trying to make a quick buck. Photography, illustration, and graphic design previously required skill or, at the very least, time to create something interesting enough to attract attention, which, online, can be converted into real money. Now free or easily affordable tools have flooded the market. What once took expert labor is now spam, powered by tools trained on the output of real artists and photographers. Spam is annoying, but ultimately easy to ignore—that is, until it collides with the negative incentives of social-media platforms, where it’s used by political shitposters and hucksters. Then the images become something else. In the hands of Trump, they create small news cycles and narratives to be debunked. In the hands of influencers, they are fired at our timelines in a scattershot approach to attract a morsel of attention. As with the Facebook AI-slop farms, social media shock jocks churning out obviously fake, low-quality images don’t care whether they’re riling up real people, boring them, or creating fodder for bots and other spammers. It is engagement for engagement’s sake. Mindlessly generated information chokes our information pathways, forcing consumers to do the work of discarding it.
That these tools should end up as the medium of choice for Trump’s political movement makes sense, too. It stands to reason that a politician who, for many years , has spun an unending series of lies into a patchwork alternate reality would gravitate toward a technology that allows one to, with a brief prompt, rewrite history so that it flatters him. Just as it seems obvious that Trump’s devoted followers—an extremely online group that has so fully embraced conspiracy theorizing and election denial that some of its members stormed the Capitol building—would delight in the bespoke memes and crude depictions of AI art. The MAGA movement has spent nine years building a coalition of conspiratorial hyper-partisans dedicated to creating a fictional information universe to cocoon themselves in. Now they can illustrate it.
Get updates delivered to you daily. Free and customizable.
It’s essential to note our commitment to transparency:
Our Terms of Use acknowledge that our services may not always be error-free, and our Community Standards emphasize our discretion in enforcing policies. As a platform hosting over 100,000 pieces of content published daily, we cannot pre-vet content, but we strive to foster a dynamic environment for free expression and robust discourse through safety guardrails of human and AI moderation.
Comments / 0