AI Slop

What AI slop is

AI slop is low-quality digital content - text, images, video or audio - churned out in bulk by generative AI tools, usually with little regard for accuracy and pushed at people who never asked to see it. The word borrows from “slop” in its older sense: the soft, formless leftovers once fed to pigs. As a label for online content, it captures something specific - not just that the material is bad, but that it was produced at industrial scale to be passed over rather than read.

The key word is volume. A single clumsy tweet is not slop; a thousand near-identical articles generated overnight to catch search traffic is. Slop is content made because getting something published cheaply mattered more than whether it was any good.

Where the term “AI slop” comes from

References to AI slop go back to at least 2022, but the word caught on in 2024 when a writer posting as “deepfates” described it as the term for unwanted AI-generated content. The developer Simon Willison then helped fix the meaning, drawing a line that has stuck: not everything an AI produces is slop, but content that is mindlessly generated and dumped on someone who did not ask for it earns the name.

By 2025 the word had gone mainstream. Merriam-Webster named “slop” its word of the year, defining it as low-quality digital content produced in quantity by artificial intelligence. When a dictionary reaches for a word to sum up the year, the thing it describes has clearly arrived.

AI slop versus useful AI-generated content

It is worth being precise here, because “AI slop” sometimes gets thrown at any work a machine had a hand in. That is not what the term is for. Plenty of AI-assisted work is careful, original and genuinely useful - drafts that a person shaped, edited and stood behind. Slop is the opposite impulse: generate fast, publish faster, and let quantity do the work that quality used to. The difference is not the tool. It is whether anyone bothered. Slop is also only one way AI output goes wrong - AI sycophancy, where a model tells you what you want to hear rather than what is accurate, is a related failure of trust.

How AI slop spreads online

AI slop spreads because the economics now favour it. Producing a passable image, article or video has dropped to almost nothing, while the rewards - clicks, ad revenue, affiliate sales, follows - stay the same. When making a hundred things costs barely more than making one, the logic of volume takes over. This is commodification applied to culture: content treated less as something to say than as units to ship.

AI slop and the attention economy

Most slop is built for the attention economy, the marketplace in which your time and focus are the product being sold. Slop producers do not need you to find their work valuable; they need you to glance, scroll or click long enough to register as engagement. That is why so much of it leans on the cheap hooks that reliably stop a thumb - outrage, sentiment, novelty - and why it so often relies on dark patterns, the design tricks that nudge you into actions you did not quite intend.

How AI slop degrades search and social feeds

The bigger problem is what slop does to the places we go for information. Mass-generated pages can climb search rankings, pushing thin, error-ridden content above the careful stuff. On social platforms, the sheer volume crowds out human creators whose work the algorithms cannot easily tell apart from the machine-made flood.

Repetition makes this worse. The more often a claim appears, the truer it tends to feel - a quirk of the mind known as the illusory truth effect. When slop repeats the same shaky assertion across thousands of pages, it borrows credibility from its own volume. Drown a question in enough noise and the truth gets harder to find at all - the same dynamic behind the firehose of falsehood.

Where you will see AI slop

AI slop on social media

The most visible slop lives on social feeds. Facebook in particular became known for surreal AI images - the “Shrimp Jesus” pictures, figures built from insects, impossible cakes - engineered purely to harvest likes and shares. Some of it is obviously fake and faintly funny. Some is convincing enough to mislead: during one US hurricane, an AI-generated image of a child clutching a puppy in a flood spread widely and was used to score political points, long after it had been shown to be fake.

AI slop in shopping, search and news

Slop has crept into more practical corners of the web too. Online shops have listed AI-written books full of invented advice, with cover art that does not match the product - one widely mocked thumbnail for the film 12 Angry Men showed nineteen smudged faces. News-shaped websites publish AI-spun articles dressed up as reporting. The result is a web where, more and more, you cannot assume a confident page was written by anyone who knew what they were talking about.

”Worksplop”: AI slop in the workplace

Slop has even picked up an office cousin, sometimes called “worksplop” - AI-generated reports, emails and summaries that look finished but turn out vague, padded or quietly wrong. Because someone downstream still has to check and fix them, slop does not always save work. Often it just moves the effort onto whoever reads it next.

Why AI slop matters

The costs of AI slop fall unevenly. Producers pay almost nothing; everyone else pays in time, attention and trust. Artists and writers see their work devalued when a machine can mimic it in seconds for free. Readers carry a new, low-grade cognitive load - the constant effort of doubting what they see. And the wider information ecosystem frays as reliable sources get buried under cheaper imitations. The decline feels of a piece with enshittification - the way platforms tend to degrade as they chase profit at their users’ expense - and feeds the unease behind the “dead internet” idea: the sense that much of what we scroll past was never made by, or for, a human at all.

It is worth keeping some perspective, though. Every leap in how we mass-produce culture has bred its share of rubbish alongside its treasures. The printing press unleashed centuries of cheap pamphlets and hack writing - the original “Grub Street” - but also widened who got to read and write. Early cinema churned out forgettable reels while building the industry that gave us everything after. AI slop is the newest sediment in a very old stream. The task is not to dam the flood but to get better at spotting what is worth keeping.

How to spot and resist AI slop

Recognising slop gets easier with practice. The visual tells are familiar by now: hands and teeth that do not add up, backgrounds that melt under a second look, text in images that dissolves into gibberish. In writing, watch for fluent but hollow prose, confident claims with no source, and an oddly even tone that never quite commits to anything. And watch for volume - a sudden wave of near-identical posts, or an account producing more than any person plausibly could.

Resisting it is partly a habit of attention. Slow down before you share. Favour sources that put a name and a track record behind their work. Treat free, frictionless, engagement-baiting content with the suspicion it has earned. Above all, keep your own judgement in play rather than handing it over - the pull to let the machine think for you is its own trap, closely tied to cognitive offloading, where convenience quietly erodes the skills we stop using. Slop is built for the scroll. The simplest defence is to stop scrolling and look properly.