AI slop (also known simply as slop) is digital content made with generative artificial intelligence that is perceived as lacking in effort, quality, or meaning, and produced in high volume as clickbait to gain advantage in the attention economy, or earn money.[1] [2] [3] [4] It is a form of synthetic media usually linked to the monetization in the creator economy of social media and online advertising.[5] Coined in the 2020s, the term has a pejorative connotation similar to spam. "Slop" was selected as the 2025 Word of the Year by both Merriam-Webster and the American Dialect Society.[6] [7]
AI slop has been variously defined as "digital clutter", "filler content prioritizing speed and quantity over substance and quality",[8] and "shoddy or unwanted AI content in social media, art, books, and search results". Jonathan Gilmore, a philosophy professor at the City University of New York, describes the material as having an "incredibly banal, realistic style" that is easy for the viewer to process.[9]
As early large language models (LLMs) and image diffusion models accelerated the creation of high-volume but low-quality text and images, discussion commenced among journalists and on social platforms for the appropriate term for the influx of material. Terms proposed included "AI garbage", "AI pollution", and "AI-generated dross". Early uses of the term "slop" as a descriptor for low-grade AI material apparently came in reaction to the release of AI image generators in 2022. Its early use has been noted among 4chan, Hacker News, and YouTube commentators as a form of in-group slang.
The British computer programmer Simon Willison is credited with being an early champion of the term "slop" in the mainstream,[10] having used it on his personal blog in May 2024.[11] However, he has said it was in use long before he began pushing for the term.
The term gained increased popularity in the second quarter of 2024 in part because of Google's use of its Gemini AI model to generate responses to search queries, and the large quantities of slop on the internet were widely criticized in media headlines during the fourth quarter of 2024. It is also used in compound terms; one such example is "Microslop", used in response to increasing integration of the Microsoft Copilot chatbot into the Windows operating system.[12] [13]
According to an academic article by Cody Kommers and five other scholars that was published in January 2026, AI slop has "so far resisted formal definition."[14] Although they argue it is impossible to precisely describe a boundary between slop and non-slop, Kommers et.al. identify three "prototypical properties" that characterise AI slop: superficial competence, asymmetric effort and mass producibility.
Beyond these family resemblances, there are many different kinds of AI slop. Three main "dimensions of variance", or ways in which AI slop can vary, are its instrumental utility (why was it created?), the level of personalization (is it so specific as to only be interesting to one person or a small friend group?) and the level of surrealism, where some AI slop is "ludicrously implausible" while other slop is more realistic.
Art curator Francesco D'Isa considers it wrong to say that AI "produces only slop", and sees the majority of human production as having "always been slop", with the canon of art being "the surviving tip of an immense iceberg of forgotten, derivative, or simply boring creations".[15]
AI image and video slop have proliferated on social media in part because it can be revenue-generating for its creators on Facebook and TikTok, with the issue affecting Facebook most notably. This incentivizes individuals from developing countries to create images that appeal to audiences in the United States which attract higher advertising rates.[16] [17] [18]
The journalist Jason Koebler speculated that the bizarre nature of some of the content may be due to the creators using Hindi, Urdu, and Vietnamese prompts (languages which are underrepresented in the model's training data), or using erratic speech-to-text methods to translate their intentions into English.[16]
Speaking to New York magazine, a Kenyan creator of slop images described giving ChatGPT prompts such as "WRITE ME 10 PROMPT picture OF JESUS WHICH WILLING BRING HIGH ENGAGEMENT ON FACEBOOK", and then feeding those created prompts into a text-to-image AI model such as Midjourney.[2]
AI-generated images of plants and plant care misinformation have proliferated on social media.[19] [20] Online retailers have used AI-generated images of flowers to sell seeds of plants that do not actually exist. Many online houseplant communities have banned AI-generated content but struggle to moderate large volumes of content posted by bots.
Facebook spammers have been reported as AI-generating images of Holocaust victims with fake stories; in reality there are only a handful of historical photographs taken at Auschwitz. The posters were described as "slop accounts", and the Auschwitz Memorial museum called the images a "dangerous distortion".[21] History-focused Facebook groups also have been inundated with AI-generated "historical" photos.[22]
Slopper, a pejorative slang term derived from "AI slop", was coined in 2025 to describe someone who is overly reliant on generative AI tools like ChatGPT.[23] [24]
Online meme content has taken on the trend of using AI-generated content to fool and entertain viewers. Users on social media have begun using AI-generated images to build massive followings by fooling viewers. This strategy has been popularized for those who are interested in an easy way to get an income online as all that is needed is one post, of the hundreds posted weekly, to gain traction and encourage the quickly made content.[25] Some creators are frustrated that their hard work is being stolen by AI-generated content. An artist, Michael Jones,[26] created physical wood carvings of animals using a chainsaw, but the style of the sculptures was taken by and used as a source for AI-generated content, which began to surface with other people beside them, claiming to have made the sculptures themselves. Jones stated that AI-slop is "a huge issue for carvers all over the world who are sadly missing out on the rightful credit exposure to their work..."[25]
According to a 2025 report by video creation and editing service company Kapwing, Korea ranks first worldwide in AI slop consumption. Experts say the phenomenon is driven by the country's rapid embrace of new technology.[27] [28]
In January 2026, YouTube CEO Neal Mohan stated that reducing slop and detecting deepfakes were priorities for YouTube in 2026.[29]
In August 2024, The Atlantic noted that AI slop was becoming associated with the political right in the United States, who were using it for shitposting and engagement farming on social media, with the technology offering "cheap, fast, on-demand fodder for content".[30] AI slop is frequently used in political campaigns in an attempt at gaining attention through content farming.[31] In 2025, in the first five months of Donald Trump's second tenure as US president, Trump posted several AI-generated images of himself on official government social media accounts, such as images of him as the pope or as a muscular man brandishing a lightsaber.[32] In August 2024, Trump posted a series of AI-generated images on his alt-tech social media platform, Truth Social, portraying fans of the pop singer Taylor Swift in "Swifties for Trump" T-shirts, as well as an AI-generated image of Swift appearing to endorse Trump's 2024 presidential campaign. The images originated from the conservative Twitter account @amuse, which posted numerous AI slop images leading up to the 2024 United States elections that were shared by other high-profile figures within the Republican Party, such as Elon Musk, who has publicly endorsed generative AI.[33] In 2025, Wired described Donald Trump as "The first AI slop President", noting his frequent use of AI-generated images and videos in public messaging. The magazine highlighted examples such as AI depictions of Trump as a fighter pilot and as a religious figure, arguing that his reliance on low-quality generative content marked a new phase in political communication.[34]
In the aftermath of Hurricane Helene in 2024, Republican influencers such as Laura Loomer circulated on social media an AI-generated image of a young girl holding a puppy in a flood, and used it as evidence of the failure of President Joe Biden to respond to the disaster.[35] [36] The Republican activist Amy Kremer shared the image while acknowledging it was not genuine.[37] [38]
The initial version of the Make Our Children Healthy Again Assessment of children's health issues, released by a commission of cabinet members and officials of the Trump administration, and led by US Department of Health and Human Services Secretary Robert F. Kennedy Jr., reportedly cited nonexistent and garbled references generated using AI.[39] [40]
In response to the No Kings protests in October 2025, Trump posted a video depicting himself flying a fighter jet and releasing feces on crowds of demonstrators, including Harry Sisson, a Democratic influencer.[41]
In the midst of disruptions to food stamp distribution during the 2025 US government shutdown, anonymous social media users began using OpenAI's Sora AI model to post slop videos of "welfare queens" complaining, stealing, and rioting in supermarkets; many comments to the videos appeared unaware that they were AI-generated, or acknowledged that they were AI-generated but nonetheless useful in pushing a narrative of widespread welfare fraud.[42] [43]
A study done by the analytics company Graphika found that the governments of Russia and China have been using AI-generated slop as propaganda. This includes the use of "spamouflage" as AI-generated content featuring fake influencers was found to be linked to China. These videos often focused on divisive topics aimed to cause disruption with ulterior motives to the presented content.[44]
In February 2025, Donald Trump shared an AI-generated video on Truth Social and Instagram depicting a hypothetical Gaza Strip after a Trump takeover. The video's creator claimed it was made as political satire.[45]
During the Gaza War, AI-generated media was used to exaggerate support for both sides, and to evoke sympathy using fake images of suffering civilians.[46] Because of content restrictions in generative AI, these images and videos rarely depict people wounded in battle, instead focusing on damage to buildings. Fake images of attacks were used to avoid accidentally providing intelligence to enemies.[47]
In November 2024, Coca-Cola used AI to create three commercials as part of their annual holiday campaign. These videos were immediately met with backlash from both casual viewers and artists;[48] animator Alex Hirsch, creator of Gravity Falls, criticized the company's decision not to employ human artists to create the commercial.[49] In response to the negative feedback, the company defended their decision to use generative AI, stating that "Coca-Cola will always remain dedicated to creating the highest level of work at the intersection of human creativity and technology".[50] Coca-Cola continued to utilize AI-generated commercials for their 2025 holiday campaign.[51]
During the holiday season of 2025, McDonald's Netherlands released an AI-generated Christmas advertisement titled It's the Most Terrible Time of the Year, which was met with a large amount of backlash. The advert was seen as cynical, portraying Christmas time as "the most terrible time of the year". The company turned off comments on YouTube and later removed the initial upload of the video from public view in response, though reuploads of the original were still public on the site.[52]
In March 2025, Paramount Pictures was criticized for using AI scripting and narration in an Instagram video promoting the film Novocaine.[53] The ad uses a robotic AI voice in a style similar to low-quality AI spam videos produced by content farms. A24 received similar backlash for releasing a series of AI-generated posters for the 2024 film Civil War. One poster appears to depict a group of soldiers in a tank-like raft preparing to fire on a large swan, an image which does not resemble the events of the film.[54] [55] In the same month, Activision posted various advertisements and posters for fake video games such as "Guitar Hero Mobile", "Crash Bandicoot: Brawl", and "Call of Duty: Zombie Defender" that were all made using generative AI on platforms such as Facebook and Instagram, which many labelled as AI slop.[56] The intention of the posts was later stated to act as a survey for interest in possible titles by the company.[57] The Italian brainrot AI trend was widely adopted by advertisers as an attempt to adjust to younger audiences.[58]
During Super Bowl LX in 2026, vodka brand Svedka aired an AI-generated commercial featuring two robots, apparently male and female, drinking Svedka while dancing at a club.[59] The ad was met with heavy criticism from Super Bowl viewers, and many took to social media to express their disdain. Svedka trained the AI on TikTok dances for the robots' dancing.
Fantastical promotional graphics for the 2024 Willy's Chocolate Experience event, which took place at Glasgow, Scotland, characterized as "AI-generated slop",[60] misled audiences into attending an event that was held in a lightly decorated warehouse. Tickets were marketed through Facebook advertisements showing AI-generated imagery, with no genuine photographs of the venue.[61]
In October 2024, thousands of people were reported to have assembled for a non-existent Halloween parade in Dublin as a result of a listing on an aggregation listings website, MySpiritHalloween.com, which used AI-generated content.[62] [63] The listing went viral on TikTok and Instagram.[64] A similar parade had previously been held in Galway. Dublin had hosted parades in prior years, although there was no parade in 2024. One analyst characterized the website, which appeared to use AI-generated staff pictures, as likely using AI "to create content quickly and cheaply where opportunities are found".[65] The site's owner said that "We asked ChatGPT to write the article for us, but it wasn't ChatGPT by itself." In the past the site had removed non-existent events when contacted by their venues, but in the case of the Dublin parade the site owner said that "no one reported that this one wasn't going to happen". MySpiritHalloween.com updated their page to say that the parade had been "canceled" when they became aware of the issue.[66]
Online booksellers and library vendors now have many titles that are written by AI and are not curated into collections by librarians. The digital media provider Hoopla, which supplies libraries with ebooks and downloadable content, has generative AI books with fictional authors and dubious quality, which cost libraries money when checked out by unsuspecting patrons.[67]
Users of Amazon Kindle and other eBook sites have reported concerns with the increased output of novels seemingly created artificially. Author Jane Friedman stated she had to report 29 novels in one week which were created by AI but used her name and likeness. Kindle created a limit of three novels per day from a single author to combat the issue.
In February 2023, the science fiction magazine Clarkesworld had to temporarily close short story submissions after receiving massive amounts of AI spam, which editor Neil Clarke attributed to people from outside the speculative fiction community trying to make easy money.[68] Clarke expressed worry that this trend would result in higher barriers of entry for new authors.[69]
As of 2024, Canadian and United States copyright laws ruled that books created by AI cannot be copyrightable. Books created by AI are mostly considered plagiarism; more specifically, non-fiction novels cannot apply for copyright due to hallucinations and biases of LLMs.[70]
includes assets generated by AI. Since the game's initial release, many players had accused Treyarch and Raven Software of using AI to create in-game assets, including loading screens, emblems, and calling cards. A particular example was a loading screen for the zombies game mode that depicted "Necroclaus", a zombified Santa Claus with six fingers on one hand, an image which also had other irregularities.[71] The previous entry in the franchise, Call of Duty: Modern Warfare III, was also accused of selling AI-generated cosmetics.[72] In February 2025, Activision disclosed Black Ops 6s usage of generative AI to comply with Valve's policies on AI-generated or assisted products on Steam. Activision states on the game's product page on Steam that "Our team uses generative AI tools to help develop some in game assets."[73] , the most recent entry in the Call of Duty franchise as of January 2026, continued the usage of AI-generated content with Studio Ghibli-styled calling cards which have been criticized by fans.[74]
In 2024, Rovio Entertainment released a demo of a mobile game called Angry Birds: Block Quest on Android. The game featured AI-generated images for loading screens and backgrounds.[75] It was heavily criticized by players, who called it shovelware and disapproved of Rovio's use of AI images.[76] [77] It was eventually discontinued and removed from the Play Store.
According to a 2025 report by AI and Games, about 20% of games published on Steam in 2025 include an "AI disclosure", indicating they use generative AI tools for some assets.[78] The same report uses explicitly the term "AI shovelware problem", arguing that many of these games rely heavily on AI-generated visuals (textures, images, dialogue, story bits, etc.), often with mixed or poor quality.
Some films have received backlash for including AI-generated content. The film Late Night with the Devil was notable for its use of AI, which some criticized as being AI slop.[79] [80] Several low-quality AI-generated images were used as interstitial title cards, with one image featuring a skeleton with inaccurate bone structure and poorly-generated fingers that appear disconnected from its hands.[81]
Some streaming services such as Amazon Prime Video have used AI to generate posters and thumbnail images in a manner that can be described as slop. A low-quality AI poster was used for the 1922 film Nosferatu, depicting Count Orlok in a way that does not resemble his look in the film.[82] A thumbnail image for 12 Angry Men on Amazon Freevee used AI to depict 19 men with smudged faces, none of whom appeared to bear any similarities to the characters in the film.[83] [84] Additionally, some viewers have noticed that many plot descriptions appear to be generated by AI, which some people have characterized as slop. One synopsis briefly listed on the site for the film Dog Day Afternoon read: "A man takes hostages at a bank in Brooklyn. Unfortunately I do not have enough information to summarize further within the provided guidelines."[85]
In one case, Deutsche Telekom removed a series from their media offer after viewers complained about the bad quality and monotonous German voice dubbing (translated from original Polish) and it was found out that it was done via AI.[86]
See also: Music and artificial intelligence and Virtual band.
In some cases, large volumes of AI-generated tracks have been uploaded with the aim of manipulating streaming platform royalties. For example, in September 2024 US musician Michael Smith was charged after using hundreds of thousands of AI-generated songs and bots to generate more than US$10 million in royalties.[87] In June 2025, Deezer estimated that as much as 70% of streams of AI-generated tracks on its platform were fraudulent, highlighting concerns about mass low quality output competing with human made music.[88]
In April 2024, Pink Floyd held a fan-made music video contest for tracks from the Dark Side of the Moon album to celebrate the band's 50th anniversary. Among ten chosen entries, the music video for the track "Any Colour You Like" was noted for being made using the Stable Diffusion software, prompting online backlash from fans against the band and the video's submitter for its AI-generated nature as well as its perceived low quality.[89] [90] [91]
In November 2024, Kanye West released an unexpected music video for the song "Bomb" from Vultures 2, featuring AI-generated versions of his daughters, North West and Chicago West. The video shows the children racing through a desert in futuristic Cybertruck-like vehicles, with both providing vocals—North partly in Japanese and Chicago in a freestyle style. The release drew widespread debate, with many fans criticizing West's increasing reliance on AI and specifically referring to the project as "AI slop", expressing discomfort with the use of AI replicas of his children.[92]
In July 2025, numerous media outlets reported on accusations that an indie band named The Velvet Sundown, which in only a few weeks had amassed over 850,000 listeners on Spotify, was AI-generated. Critics observed that there were no records of any performances by the band, that individual band members had no social media presence, and that their promotional images appeared to be fake.[93] Deezer's AI detection tool flagged the band's music as being 100% AI-generated.[94] Rolling Stone (which described The Velvet Sundown as "obviously fictional") had reported that a spokesperson for the band named Andrew Frelon had admitted that their music was an "art hoax" generated using the AI tool Suno. However, Frelon later stated that his statement was itself a hoax and that he had no connection to the band.[95] However, within a week, the Velvet Sundown's artist biography on Spotify had been updated to say that the band was "a synthetic music project guided by human creative direction, and composed, voiced, and visualized with the support of artificial intelligence", intended as "artistic provocation".[96] [97] By this time, the band had over one million monthly listeners on Spotify.[98] According to a former Spotify employee, the high listener count has two likely causes: firstly, Spotify now accepts payments to boost playlist placement, and secondly, playlists are increasingly selected by algorithms rather than humans.[99]
AI has also been used to impersonate established musicians, and release music on major streaming services under their name without their knowledge, presumably to make money from streaming royalties. In particular, in August 2025, a number of Americana and folk-rock musicians including Jeff Tweedy, Father John Misty, and Blaze Foley (who died in 1989) were impersonated, as well as some established US Christian musicians and metalcore bands. The fake releases all had similar AI-generated cover art, and were credited to the same three record labels. Many listed "Zyan Maliq Mahardika" as a songwriter, indicating that the impersonation has a single source. Spotify removed the tracks, stating that they "violated our policy against impersonating another person or brand".[100]
In November 2025, the AI-generated song "Walk My Walk" topped the Billboard's country digital song sales chart in the United States with a mere 3,000 sales.[101] The song is by AI artist Breaking Rust, who amassed over 4.5 million Spotify listeners on their songs. Promotion for "Walk My Walk" was accompanied by an AI slop video shared on Instagram depicting a cowboy walking into the sunset.[102] In the same month, an AI-generated song named "We Are Charlie Kirk", made in tribute to the assassination of Charlie Kirk two months earlier, went viral on TikTok and ranked #1 on Spotify's viral songs chart.[103] That month in the Netherlands, the AI-generated far-right anti-immigrant protest song "Wij zeggen nee, nee, nee, tegen een AZC" (We say no, no, no to an asylum shelter) peaked at No. 5 on the Dutch Single Top 100.[104]
Generative AI has been used to write articles which have been published in both low-quality research paper mills and reputable scientific journals.[105] In 2024, a peer-reviewed article containing a generated image of a rat with absurdly large genitals accompanied by nonsensical text and diagrams was retracted by Frontiers in Cell and Developmental Biology after drawing attention from scientists on social media.[106] [107]
Likewise, in 2025, Nature's Scientific Reports retracted the peer-reviewed article Bridging the gap: explainable ai for autism diagnosis and parental support with TabPFNMix and SHAP because of 'an AI-generated figure' and other significant errors.[108]
David Berry has used the term scholarslop[109] [110] to refer to AI-generated administrative discourse in universities, or quasi-academic texts. Glen Berman argues that AI-generated articles being published in scholarly journals are an "epistemic carcinogen" that will pose a major risk to the knowledge ecosystem.[111]
In January 2026, the open-source command-line utility cURL ended its bug bounty program through HackerOne after receiving a large number of false vulnerability reports.[112] In an announcement on its official GitHub page, the project stated that rewards would no longer be offered for reported vulnerabilities. Daniel Stenberg, founder and lead developer of cURL, shared examples of fake, AI-generated bug reports submitted through the program, including one that claimed a critical vulnerability that did not actually exist. Many of these reports referenced nonexistent changelogs, included code snippets that did not match real function signatures, or were the result of simple user errors. Stenberg explained that the influx of AI-generated reports placed a significant burden on cURL's security team and that ending the program was an "attempt to reduce the noise."[113]
A Harvard Business Review study, done in conjunction with Stanford University and BetterUp, found that employees were using AI tools to create low-effort "workslop" that created more work for their colleagues.[114] Within the timeframe of the study, it was found that 40% of participating employees received some form of "workslop", with each incident taking an average of two hours to resolve. BetterUp defines workslop as "AI-generated content that looks good, but lacks substance".[115] The study appears to be the first instance of "workslop" and also uses "workslopped" as a verb.[116]
An AI-generated scene of Sabrina Impacciatore skating and skiing across historic host cities at the 2026 Winter Olympics opening ceremony was criticized for having replaced animators.[117] As well as a poor depiction of the Olympic rings, violating the Olympic brand guidelines of the Olympic symbol, posted by the official Olympics account.[118] The ice dancers Kateřina Mrázková and Daniel Mrázek were also criticized for dancing to an AI-generated song that combined AC/DC's "Thunderstruck" with the AI-composed track "One Two", imitating Bon Jovi's melody.[119]
