is it slop? Check a passage

What is AI slop?

AI slop is content a machine generated with little or no human effort, made in bulk because bulk is cheap. It covers text, images, video and audio, and is also called slop content, or just slop. The word judges as much as it describes: slop is usually not worth your time, and often wrong, because a model predicting the next likely word predicts straight through facts it does not have. Merriam-Webster made slop its word of the year for 2025, defining it as low-quality digital content usually produced in quantity by artificial intelligence. What separates slop from bad writing is not how bad it is: nobody sat with it, nobody checked it, and there is plenty more where it came from.

An illustrated robot at a laptop, pencil in hand, with notebooks and loose paper spread across the desk.

Where the word came from

Slop is an old word for kitchen waste and pig feed, so the insult was built in. Wikipedia's article on AI slop traces the first uses to around 2022, as slang on 4chan, Hacker News and YouTube for the flood of pictures from new image generators. It became the general word two years later. The writer known as deepfates noted on X that slop was becoming a term of art the way spam had for unwanted email, and the programmer Simon Willison quoted that in a May 2024 post arguing slop was the ideal name. Neither coined it. Willison's test holds: if something is "mindlessly generated and thrust upon someone who didn't ask for it", it is slop.

Merriam-Webster chose slop in December 2025, and the American Dialect Society voted it word of the year in January 2026, noting that AI slop had been a nominee the year before and that by 2025 the bare word stood on its own, the AI part understood.

What AI slop looks like

Graphite sampled tens of thousands of English-language Common Crawl pages and estimated that AI-generated articles overtook human-written ones in November 2024. Take that as an estimate rather than a census: the classifier was an AI detector with a measured false positive rate of about 4%, and the same analysis found they largely do not surface in Google.

Documented cases:

Nobody was at the wheel. Every case had someone who could have read it before it went out and did not. That absence is what the word points at.

YouTube names it too. Its monetisation rules now set out three kinds of inauthentic content, one being generic template-based video with little variation between uploads. Channels that lean on it lose their share of ad revenue.

The AI slop meme

Two things go by that name. The first is slop that is funny on purpose. Italian brainrot began on TikTok in January 2025 with a shark in sneakers called Tralalero Tralala and grew into a bestiary of AI-generated creatures with pseudo-Italian names. The joke and the slop are the same object; nobody is fooled.

The second is the word itself, now a one-word verdict left under anything people suspect a machine made. Human writing gets called slop for tidy structure, formal vocabulary, or a fondness for the em dash. The accusation costs nothing to make and a great deal to answer, so be sure first.

Is all AI writing slop?

No. Slop is a claim about effort as much as origin. If someone drafted with a model, rewrote it, checked the facts and put their name on it, that is their writing, and calling it slop is an insult. What makes slop slop is that none of that happened: the output went out as it arrived, at volume, with nobody accountable for whether it was true. Which is why "was this written by AI?" and "is this any good?" are separate questions. Machine-written text can be accurate, human-written text can be worthless, and a detector answers only the first.

How do you spot AI slop?

You spot it by checking the claims rather than the style. Slop comes out of a system built to produce plausible text, so it is fluent by construction and unreliable exactly where it needs to know something: names, dates, titles, quotes, citations, prices. Pick one checkable fact and check it. That catches more slop than any list of stylistic tells, and accuses nobody of anything.

The stylistic tells are real, and none settles anything alone. Paragraphs of even length, stock vocabulary, heavy punctuation and confident specifics with nothing behind them describe house-style corporate copy, a press release, a first draft, or someone who learned formal English as a second language just as well. That is why a checklist of tells is the wrong tool for the job. Each tell, and where it goes wrong, is on signs of AI writing.

Checking a passage with an AI slop detector

An AI slop detector is a classifier: it weighs a passage against the statistical habits of machine-generated text and returns a likelihood that a model wrote it. It does not know who typed the words, and it is not evidence.

The limits here are documented by the people who build these tools. OpenAI withdrew its own AI text classifier in July 2023, citing a low rate of accuracy. A 2023 study by Stanford researchers found detectors consistently misclassified non-native English speakers' writing as machine-generated, and that simple rewriting instructions could both remove that bias and slip text past them. Detectors are weakest on short passages, where there is too little text to judge. Nobody in this field can honestly offer you certainty. Those limits bite hardest in a classroom, where a score can start a misconduct case, so the AI detectors schools and colleges use have their own page, on what a flag from one is worth.

The detector here has the same limits. It shows which parts of a passage look machine-made, because a number with no reasoning attached is not something you can argue with. For now it reads text; images, files and links are coming.

A high score is a reason to ask, not an answer. Calling human writing machine-written is the expensive mistake, so read the flagged passages before judging the writer.

If something reads like slop, check one fact in it first: that costs a minute and settles most cases. If it still reads like slop, paste it in and read the reasons the check gives you rather than the number on its own.

Check a passage →

Last updated 2026-08-17.