The Year in Slop | The New Yorker

The Turing Test, a long-established tool for measuring machine intelligence, measures the point at which a text-generating machine can fool a human into thinking it is not a robot. ChatGPT passed this test earlier this year, ushering in a new technological era, although not necessarily one of superhuman intelligence. However, more recently, artificial intelligence has passed another threshold, a kind of Turing test for the eye: the images and videos that AI can create are now sometimes indistinguishable from the real thing. As companies like OpenAI, Meta, and Google trained, refined, and released new, image-friendly models, the online public was able to instantly create realistic AI content on any topic they could imagine, from fan art about superheroes and cute animals to scenes of violence and war. “Sucks,” a (non)art term for content produced by AI, became ubiquitous in 2025, inspiring new sub-coins such as “suck,” a pejorative term for those who rely on AI to think for itself. Slop went beyond surreal entertainment or frivolous entertainment; Gone are the relatively carefree days of bizarre and obviously fake Jesus Shrimp images on Facebook feeds. In 2025, the President of the United States used the artificial intelligence “agitation slop” to promote his policies and mock his detractors, and other politicians followed suit. Sam Altman, OpenAI's CEO, has become something of a ubiquitous mascot for Sora, his own company's social media trash feed. Not all of the content was compelling, but most of it came close enough – and in our increasingly audiovisual digital world, that may prove to be a more meaningful Rubicon than the Turing Test.

Studio Ghibli Factory by OpenAI

Looking back, it seems almost strange. Towards the beginning of the year, OpenAI released an updated GPT-4o model that had the ability to generate still images in a ChatGPT text window. One of the demos used to promote the product was a group selfie of Sam Altman and a couple of his employees turned into a drawing. in style legendary Japanese animator Hayao Miyazaki, animation house Studio Ghibli. The model emulated Studio Ghibli's organic, highly human aesthetic amazingly well; the stunt went viral and the internet was flooded with Ghibli slop consisting of random images easily turned into anime. A troubling low was reached when White House account X cynically used the meme to promote a roundup of immigrants, sharing a Ghibli image showing the arrest of a black woman described as a “felon alien convicted of trafficking fentanyl.”

The Bible of Artificial Intelligence

In May, Google released Veo 3, an artificial intelligence model for video generation that can play eight-second photorealistic clips. A couple of months later, an organization called AI Bible published eight minute video on YouTube, where Rapture was rendered using artificial intelligence reminiscent of clips from Game of Thrones. No prestige TV budget was required, just the patience to type clues into a text box. This slopocalypse has received over eight hundred thousand views, indicating how readily online audiences will embrace the empty facsimiles as must-see TV shows.

Animals jump on trampolines

AI, like an audiovisual genie, can grant wishes by depicting fantastical things we would like to believe are real and testing our willingness to suspend disbelief. In June, a California man captured real-life video of coyotes wandering around a trampoline and playfully bouncing. The footage sparked a frenzy of AI videos of other animals jumping on trampolines in ways that would surely defy verisimilitude if they weren't so adorable: rabbits jumping taller than usual, moose performing a somersaultbears fuck until the mat fails. Some of these shots are considered real in part because they use the aesthetic of infrared surveillance cameras, showing dark creatures at night. Those paying close attention may have noticed that animals tend to appear out of nowhere, multiply, or disappear, but of course, when you're watching the video, there's little incentive to look twice.

Sam Altman steals GPUs

At first, slop was a widely derided format, the kind of clickable nonsense churned out by content farms or trolls. But in September, with the launch of OpenAI's Sora app, social networks and feed For AI-generated videos, the company tried to convince users that slop is something we should all do for fun. Sora allowed you to become a sort of artificial intelligence puppet that other users could adopt and remake as they wished. Altman became a test subject for the feature, resulting in one of the most popular videos on the new platform: a clip in which Altman steals GPUs (the graphics processing units needed to run artificial intelligence) from shelves at a Target store and is apprehended by a security guard. The video was intended as a joke, but the convincing fake photo of a man committing a crime is exactly what is meant to scare us.

Slop Andrew Cuomo vs Mamdani

AI's new talent for creating realistic videos was quickly used as a political weapon. In October, former Gov. Andrew Cuomo's mayoral campaign released an AI-generated ad featuring “Criminals for Zoran Mamdani,” a cast of characters ranging from a wife beater to a shoplifter based on crude and often racist stereotypes, all of whom support the now-elected mayor. The video, which the campaign quickly deleted, demonstrated AI's uncanny ability to create realism devoid of ethics; it's hard to imagine human actors agreeing to collaborate with such a ridiculously offensive script. Shortly after the ad's release, accounts titled “Citizens Against Mamdani” appeared on social media, showing supposed locals loudly berating the “stupid” voters who carried Mamdani to victory. This apparent attempt by AstroTurf to create a movement against Mamdani was thwarted when some accounts were flagged as AI and removed. These pseudo-New Yorkers were frighteningly realistic, even more so than Zoran's criminals.

Trump's flying excrement and MedBed slop

Donald Trump kicked off the year by sharing an AI video of “Trump Gaza,” a reconstruction of the Palestinian territory's cityscape dotted with golden statues of him. It was created two Israeli-American directors ridiculed Trump's idea of ​​turning Gaza into the “Riviera of the Middle East,” but for Trump it was simply a useful mock-up of his vision. The president's budding love affair with AI videos came into full bloom in October when he shared a generated video of himself piloting a bomber with the caption “King Trump” on it and dropping what appeared to be feces on “No Kings” protesters. The video, relatively cartoonish but disgusting nonetheless, would likely have remained a strange internet ephemera had it not been posted by the real commander-in-chief, presumably as a demonstration of what he would like to do to his opponents. Another video Trump shared around the same time was featured on Lara Trump's Fox News program, which praised hospitals specializing in an alleged miracle technology based on a conspiracy theory known as “MedBeds.” Either Trump was somehow unaware that the footage was fabricated, or he deliberately chose to promote it nonetheless. Either way, it was deepfake news.

Mural “Christmas slop” on the Thames

As slop becomes easier to create, it moves away from our screens and into physical spaces. In November, residents of the London suburb of Kingston upon Thames were shocked to discover that a huge festive mural installed in a city center shopping area combined Christmas cheer with artificial intelligence horror. The scenes were supposed to depict a winter fair on a frozen river; the participants, however, were mutant dog-bird hybrids, snowmen with too many eyes, and humanoid bodies stuck in an impossibly thick quagmire. Rumor has it that the murals were created by British artist Matt Collishaw at the request of a local homeowner. The murals sparked political controversy (were they some kind of veiled commentary on immigration?) and were soon torn down. However, one city resident praised the work, calling it “so bad it's good.”

McDonalds Pure AI Advertising

If 2025 marked the advent of slop, it also brought the backlash that came with it. The shallowness, glitches, and overly smooth textures of AI content have become symbols of cunning mixed with laziness. This month, McDonald's in the Netherlands released an entirely AI-generated holiday ad called “It's the Worst Time of the Year,” depicting various holiday commotions: Christmas trees toppling over, baking disasters, carolers caught in a snowstorm. The solution, according to the ad, is to go into a warm, cozy, fake McDonald's restaurant and hide until January. Both because of the negative attitude towards Yuletide rituals and because of a pathetic attempt to save on production costs, the ad was so poorly received that the company decided to abandon it. McDonald's Netherlands has apologized, admitting that for many of its customers the holidays are actually “the most wonderful time of the year.” Nobody wants to find slop under the tree. ♦