AI Gone Wild: How Generative AI is Turning the Web into a Digital Dumping Ground

What is AI Slop?

“AI slop” refers to digital content produced by generative AI that is characterized by its lack of effort, poor quality, and overwhelming volume of production. This term encompasses images with glaring anatomical errors, texts riddled with inconsistencies, surreal videos, and fabricated reviews.

Notable examples include “Shrimp Jesus” (a depiction of Jesus covered with shrimp), images of Donald Trump dressed as the pope, and a scientific publication withdrawn by the journal Frontiers in Cell and Developmental Biology that featured an anatomical image of a rat with absurdly oversized genitalia.

“The term ‘slop’ can be translated as ‘mush,’ ‘trash,’ or ‘nonsense.’ Yet, we see it everywhere,” explains Albertine Meunier, an artist and author of the “Slop Machine,” a work that comments live on streams of slop on Instagram to Libération.

The Origin of the Term “AI Slop”

The expression originated in online forums such as 4chan, Hacker News, and YouTube comments, where it emerged in 2022 in response to the release of the first AI image generators. Initially, “slop” referred to the sludge that accumulates at the bottom of oil tanker holds.

It was British developer Simon Willison who popularized the term for AI in May 2024 on his personal blog. The term then skyrocketed in popularity, especially with Google’s deployment of an image generator in Gemini, which was notably “sloppy.” The term was officially recognized at the end of 2024 when Merriam-Webster named “slop” the word of the year for 2025.

A derivative term, “slopper,” emerged to pejoratively describe a person overly reliant on generative AI tools and mass-producing such content (any resemblance to the 47th President of the United States is coincidental).

Why Slop is Exploding on the Web

A Not-So-New Phenomenon

This phenomenon emerged in 2022 with the democratization of image (Midjourney, Stable Diffusion) and text (ChatGPT, Gemini) generation models. “Creating an image has become easy, the tools are now accessible to everyone,” notes Albertine Meunier in Libération.

The main reason for its proliferation is economic. Platforms like TikTok, Instagram, and Facebook pay for popular content. A case documented by the Washington Post shows that a video of a kangaroo in an airport made $15,000 in three months.

The Easy Money of Slop

This attention economy particularly attracts creators from developing countries who produce content targeting Western audiences, where advertising rates are higher. The New York Magazine, for example, interviewed a Kenyan creator who used ChatGPT to churn out images “that will bring a significant engagement rate on Facebook.”

“Slop is designed to capture attention, to stand out in the continuous flow of our news feeds,” analyzes Valentina Tanni, a meme historian, for the French daily. “That’s why videos often focus on the absurd, the nonsensical, the unexpected, the shocking, or even the disgusting.”

Real-life examples, at least those causing reactions, are plentiful: the band The Velvet Sundown accumulated 850,000 listeners on Spotify before being revealed as entirely AI-generated. In Dublin, thousands gathered for a Halloween parade that never existed, listed by a site using AI content…

The Dangers of Slop According to Experts

Experts are sounding the alarm. Melanie Mitchell, an American AI researcher, points out that AI systems are not sophisticated enough to distinguish real from fake.

A study by Harvard Business Review reveals that 40% of employees receive “workslop,” AI-generated content that “looks good, but lacks substance” and requires an average of two hours to correct. On Deezer, up to 70% of AI-generated song streams are fraudulent.

“For me, it’s a stage beyond deepfake, because the more you watch them, the more detached from reality you become,” warns Albertine Meunier. Jason Koebler, a journalist for 404 Media, discusses, “beyond this profusion of low-quality content, which is easily adaptable to a platform’s performance at a given moment, the near-collapse of the informational ecosystem and, consequently, of ‘reality’ online.”

The political dimension of slop also causes concern. AI specialist Kate Crawford recalls a statement by far-right propagandist Steve Bannon in 2020: “The real opposition is the media. And the best way to fight them is to flood them with shit.” Unfortunately, he does it all too well…

Between Threat and Cultural Phenomenon

In light of these concerns, some well-known figures are trying to nuance a debate in which they could be implicated. Two weeks after Merriam-Webster named “slop” the word of the year, Satya Nadella, CEO of Microsoft, published a post on his personal blog. “We need to move beyond the arguments that pit slop against sophistication and develop a new concept that evolves the idea of ‘bicycles for the mind’ so that we always think of AI as a scaffold for human potential rather than a substitute,” wrote Satya Nadella.

The debate between replacing and augmenting human capabilities continues. MIT’s Project Iceberg estimates that AI can perform about 11.7% of paid human work. Paradoxically, the 2026 Vanguard report shows that “the approximately 100 professions most exposed to automation by AI outperform the rest of the job market in terms of employment growth and real wage increases.”

Slop is also becoming a cultural object. The exhibition “From Spam to Slop,” held in late 2025 in Paris, explored this phenomenon. Artists like Bennett Waisbren, whose video reached a billion views on Instagram, and Robin Lopvet question the notion of authorship: “I don’t work on my own images, I just animate found images. And ultimately, it’s the machine that does it. What is the author’s role in these creations?”

Meanwhile, platforms are beginning to respond: Kindle has imposed a limit of three novels per day per author, Spotify removes fraudulent content, Deezer has developed an AI detection tool…

Slop fundamentally questions the value of content in the era of AI and the future of online information. It reveals the tensions of the attention economy in an increasingly machine-generated web.