What Is AI Slop?
AI slop is low-quality, templated content that sounds fluent but says nothing. It's Merriam-Webster's 2025 Word of the Year, and it's flooding the internet at an unprecedented scale.
Table of Contents
What AI Slop Means
AI slop (noun): Low-quality digital content produced in quantity by artificial intelligence that sounds fluent but communicates nothing of substance.
"Digital content of low quality that is produced usually in quantity by means of artificial intelligence."
— Merriam-Webster, Word of the Year 2025 (source)
The critical distinction: AI slop is about quality patterns, not origin. Not all AI content is slop — and not all slop is AI-generated. Humans have been writing empty, templated filler for decades. But generative AI has made it possible to produce this kind of content at a scale never seen before.
Researchers have identified three prototypical properties of AI slop:
Superficial Competence
Sounds fluent, polished, even authoritative — but communicates nothing of substance under scrutiny
Asymmetric Effort
Costs almost nothing to produce but demands real time and attention from readers to evaluate
Mass Producibility
The existence of any slop implies an endless supply nearby. There is no artisanal, hand-crafted slop
In practice, AI slop exhibits these recognizable patterns:
Templated
Follows predictable patterns and structures: "In today's fast-paced world..."
Vague
Lacks specific details, concrete examples, or actionable information
Performatively Insightful
Sounds profound but communicates nothing new: "The key is to find balance..."
No Lived Experience
Contains no personal perspective, no stakes, no specificity that comes from actually doing the thing
As programmer Simon Willison put it: sharing unreviewed AI-generated content with other people is "rude." His principle is simple: "Don't publish slop."
Origin & History of "AI Slop"
The term "slop" has a long history. It first appeared in the 1700s meaning "soft mud," evolved to "food waste" in the 1800s, and eventually came to mean "rubbish" or "a product of little or no value." In the 2020s, internet communities repurposed it for AI-generated content — similar to how "spam" became the word for unwanted email.
Early uses of "slop" as slang for AI-generated content appear on 4chan, Hacker News, and YouTube — initially referring to AI image generators.
Programmer Simon Willison championed the term in a viral blog post, arguing that "slop" should become the standard word for unwanted AI content — just as "spam" did for unwanted email. He cited Microsoft's AI travel guide recommending the Ottawa Food Bank as a tourist dining destination as a perfect example.
Source: simonwillison.net
Usage surged as Facebook was flooded with AI-generated images (the infamous "Shrimp Jesus" phenomenon). During Hurricane Helene (September 2024), AI-generated images of disaster victims went viral, including a fabricated image of a crying girl in a boat — even a U.S. senator shared one before realizing it was fake.
John Oliver dedicated a full Last Week Tonight segment to AI slop, bringing the concept to mainstream audiences.
Source: HBO
Australia's Macquarie Dictionary named "AI slop" its Word of the Year 2025 — winning both Committee's Choice and People's Choice (only the fourth time both aligned).
Source: Macquarie Dictionary
Merriam-Webster named "slop" its Word of the Year 2025. The American Dialect Society followed suit in January 2026, voting it their Word of the Year too — over 300 linguists participated.
Source: Merriam-Webster
YouTube CEO Neal Mohan declared "managing AI slop" a top priority for 2026, announcing expanded deepfake detection and AI content labeling. Over 1 million YouTube channels were using AI tools daily.
Source: CNBC
"Slop": The Word of the Year 2025
In December 2025, Merriam-Webster announced "slop" as its Word of the Year. It was a remarkable cultural moment — a slang term born on niche internet forums had become the defining word of an era.
"Absurd videos, off-kilter advertising images, cheesy propaganda, fake news that looks pretty real, junky AI-written books, 'workslop' reports that waste coworkers' time... and lots of talking cats."
— Merriam-Webster's examples of AI slop (source)
Australia's Macquarie Dictionary had named "AI slop" its Word of the Year a month earlier. Then in January 2026, the American Dialect Society voted it their Word of the Year too — making it a rare triple-crown: three major language authorities all converging on the same word.
The ADS noted that while "AI slop" was already a nominee in their 2024 vote, by 2025 "slop" could stand on its own — the AI context was implicitly understood. The word had transcended its origins to become a cultural shorthand.
How Big Is the AI Slop Problem?
The numbers are staggering — and growing fast:
of newly published online articles are now AI-generated, up from ~10% in late 2022 — the split has plateaued at roughly 50/50
Source: Graphite / Futurism, Oct 2025
of 900,000 newly created web pages contained AI-generated content (only 25.8% were purely human-written)
Source: Ahrefs, April 2025
of YouTube Shorts shown to new users are AI slop, with an additional 33% qualifying as 'brainrot'
Source: Kapwing, Nov 2025
undisclosed AI-generated news websites identified, spanning 16 languages
Source: NewsGuard
reduction in low-quality search results after Google's March 2024 core update targeting AI slop
Source: Google Search Blog
A key insight from Graphite's research: while 52% of new articles online are AI-generated, only 14% of articles appearing in Google Search results are AI-written — meaning Google's algorithms are effectively filtering out most AI slop from search results, even as it floods the web.
On Amazon, an estimated 10,000-40,000 AI-generated e-books are released each month — many without disclosure. Some are dangerous: AI-written mushroom foraging guides with potentially lethal misidentifications have been found on Amazon. And at Clarkesworld, the award-winning science fiction magazine, AI submissions overwhelmed their slush pile in 2023, forcing them to shut down submissions entirely before developing AI detection tools.
Real-World Examples of AI Slop
AI slop isn't just text — it's images, videos, books, and entire websites. Here are some of the most notorious examples:
Shrimp Jesus (Facebook, 2024)
Surreal AI-generated images of Jesus fused with shrimp, posted with engagement-bait captions like 'Say Amen for 7 years of luck.' Stanford Internet Observatory studied 120 Facebook Pages posting AI-generated images that generated hundreds of millions of interactions. The pages — primarily operated from India and Vietnam — used sensational AI images to drive traffic to ad-laden websites.
Source: Stanford Internet Observatory
Hurricane Helene Fake Images (2024)
During Hurricane Helene (September 2024), AI-generated images of disaster victims went viral on Facebook — including a fabricated image of a crying girl in a boat clutching a puppy. U.S. Senator Mike Lee shared one before realizing it was fake. The images undermined disaster relief efforts and spawned fraudulent donation schemes.
Source: Virginia Tech / PetaPixel
AI News Anchors (YouTube, 2024-2025)
Deepfake 'reporters' delivering fabricated news about celebrities, trials, and current events. Nine of YouTube's top 100 fastest-growing channels featured AI-generated content. Some AI slop channels accumulated billions of views and millions in ad revenue before being taken down.
Amazon AI Books (2024)
An estimated 10,000-40,000 AI-generated e-books published monthly on Amazon, including fake travel guides with invented restaurants, AI-written mushroom foraging guides with potentially lethal misidentifications, and plagiarized content repackaged as original works.
Source: Futurism
Clarkesworld Magazine (2023-2025)
The award-winning sci-fi magazine was forced to shut down submissions in February 2023 after being overwhelmed by AI-generated stories — 500+ AI submissions in under 20 days, driven by 'make money with ChatGPT' schemes promoted on YouTube and TikTok. They've since developed AI detection tools but still receive ~1,200 submissions monthly.
Source: neil-clarke.com
Slopsquatting (2025)
A cybersecurity attack where hackers exploit AI code generators that 'hallucinate' fake software package names. Attackers register these phantom packages with malicious code, then wait for developers using AI tools to install them. Research found 43% of hallucinated package names recur consistently, making them reliably exploitable.
Source: Infosecurity Magazine
Want to see more examples?
Browse the AI Slop Examples Gallery →The 5 Types of AI Slop (Slop Taxonomy)
Not all slop is the same. We categorize it into five recognizable patterns. Each type has distinct signals you can learn to spot:
Generic Slop
Templated, vague, no concrete details. The verbal equivalent of stock photos.
Signals:
- •Opens with broad generalizations
- •Uses filler phrases like 'In today's world...'
- •Could apply to almost any topic
- •No specific names, dates, numbers, or examples
"In today's fast-paced world, it's more important than ever to stay ahead of the curve. Whether you're a beginner or an expert, understanding the fundamentals is key to success."
Pseudo-Insight Slop
Sounds deep but delivers zero information gain. Philosophy for people who don't read philosophy.
Signals:
- •Profound-sounding but content-free statements
- •Obvious observations dressed up as wisdom
- •Vague metaphors that don't clarify anything
- •Tautologies presented as insights
"The journey of a thousand miles begins with a single step. Remember, it's not about the destination — it's about the journey. When you change the way you look at things, the things you look at change."
Fake Authority Slop
Confident expert tone without substance or citations. Trust me, bro — in article form.
Signals:
- •"Studies have shown..." (no citation)
- •"Experts agree..." (which experts?)
- •Authoritative voice with no credentials shown
- •Claims presented as facts without evidence
"Research has consistently demonstrated that implementing these strategies leads to significant improvements. Industry leaders have long recognized the importance of this approach, and the data speaks for itself."
Wikipedia Rehash
Rephrased encyclopedia definitions. Information you could get from the first paragraph of any Wikipedia article.
Signals:
- •"X is defined as..."
- •Basic definitions presented as original content
- •No analysis beyond what's commonly known
- •Educational tone without educational value
"Machine learning is a subset of artificial intelligence that enables computers to learn from data. It has many applications in various fields including healthcare, finance, and transportation."
Wellness Slop
Safe, soothing, universalized self-help. The content equivalent of a live, laugh, love sign.
Signals:
- •Gentle, non-confrontational advice
- •Universalized statements that apply to everyone
- •"Remember to be kind to yourself..."
- •Wellness buzzwords without actionable guidance
"Remember that self-care isn't selfish — it's necessary. Take time to breathe, practice gratitude, and honor your journey. You are exactly where you need to be."
Deep dive into each type with more examples and detection tips
Explore the Full Slop Taxonomy →AI Slop Detection vs AI Detection
This is the most important distinction to understand:
AI Detectors
"Was this written by AI?"
- —Judge origin, not quality
- —Often inaccurate with false positives
- —Conflate "AI-written" with "bad"
- —Output: probability percentage
AI Slop Detector
"Is this worth reading?"
- +Judge quality and substance
- +Human-readable explanations
- +Recognizes AI can produce good content
- +Output: judgment + evidence + taxonomy
Why AI Slop Matters
1. It Degrades Information Quality
When 74% of new web pages contain AI-generated content, the signal-to-noise ratio collapses. Valuable information from genuine experts gets buried under an avalanche of content-free filler. Search results become less useful, and people waste more time finding answers.
2. It Erodes Trust
When readers can't tell if content was crafted with care or mass-produced in seconds, they disengage from all content. This hurts genuine writers, researchers, and creators who invest real effort.
3. It Can Be Dangerous
AI-generated mushroom foraging guides with lethal misidentifications. Fake medical advice. Fabricated legal information. AI slop isn't just annoying — when it enters high-stakes domains, it can cause real harm.
4. It Harms Mental Health
Researcher Laura Glitson describes the "slop-doom feedback loop" — where exposure to AI slop produces feelings of "surreality, paranoia, suspicion, and anxiety." As one Reddit user put it: "The internet makes me miserable for 80% of the time I'm on it, but I just can't get out." The uncanny quality of slop creates what psychologists call the "black box effect" — unease from not understanding what's real anymore.
5. It's Invading the Workplace ("Workslop")
"Workslop" — AI-generated workplace content — is becoming its own problem. Generic emails, AI-padded performance reviews full of "synergy" and "leveraging core competencies," recruiter messages that match job postings suspiciously well, and marketing copy that is polished but immediately forgettable. It creates what Fisher Phillips calls "productivity theater" — high output that masks downstream cleanup work.
6. It's a Cultural Phenomenon
"Slop" became the Word of the Year across three major language authorities for a reason. People intuitively recognize it even when they can't articulate why. The word has spawned an entire vocabulary: slopocalypse, slopwashing, slopfluencer, sloptimization. A society doesn't invent 20+ words for something unless it's deeply felt.
What Makes Content NOT Slop?
The opposite of slop isn't "human-written" — it's content with substance:
Specific Details
Names, dates, numbers, concrete examples that anchor claims in reality
Lived Experience
Personal perspective, real stakes, specificity that comes from actually doing the thing
New Information
Something you didn't know before reading — not just rephrased common knowledge
Clear Point of View
Takes a stance, makes an argument, has a perspective that not everyone would agree with
How Platforms Are Responding
Google's March 2024 core update specifically targeted AI slop, integrating the helpful content system into its core algorithm. The result: a 45% reduction in low-quality, unoriginal content in search results — exceeding their initial 40% target. Google's stance: they don't penalize content for being AI-generated, but they do penalize content for being unhelpful.
Source: Google Search Blog
Meta / Facebook
Meta began labeling AI-generated content in May 2024. By October 2024, over 360 million pieces of content had been labeled on Facebook. In 2025, Meta began disallowing monetization for repetitive, unoriginal AI content.
Source: Meta Transparency Center
YouTube
In January 2026, YouTube CEO Neal Mohan declared "managing AI slop" a top priority for the year. YouTube now requires creators to disclose AI-generated content, labels AI-produced videos, and is expanding its "likeness detection" system — which flags when a creator's face is used without permission in deepfakes — to millions of creators. Over 1 million channels were using YouTube's AI tools daily by December 2025.
Source: CNBC
The Slop Vocabulary
A linguistic analysis by Making Science Public found that "slop" has spawned an entire vocabulary of derived terms — a sign of how deeply the concept has penetrated culture. A society doesn't invent 20+ words for something unless it's a real, daily experience:
Workslop
AI-generated workplace content: generic emails, padded reports, hollow performance reviews
Slopocalypse
The anticipated collapse of internet content quality under AI-generated flood
Slopwashing
Disguising AI content as human-created, like greenwashing for content authenticity
Slopsquatting
Cybersecurity attack exploiting AI code hallucinations to plant malicious packages
Slopfluencer
Social media influencer producing LLM-like, templated content at scale
Sloptimization
Optimization that produces degraded outputs — optimizing for AI generation, not quality
Slopper
Someone who overrelies on AI tools without review (also: 'botlicker', 'second-hand thinker')
Slopaganda
AI-generated propaganda and political messaging at industrial scale
The word "slop" itself has also become a verb: "Everyone is incentivized to slop as hard as possible."
How to Detect AI Slop
You can spot AI slop by looking for these red flags:
Or just use SlopDetector
Paste any text and get an instant verdict with evidence highlighting and human-readable explanation.
Try the Detector — Free →Frequently Asked Questions
What does AI slop mean?
AI slop refers to low-quality, templated content generated by AI that sounds fluent but communicates nothing of substance. Merriam-Webster, the American Dialect Society, and Australia's Macquarie Dictionary all named it Word of the Year for 2025, defining it as "digital content of low quality that is produced usually in quantity by means of artificial intelligence."
Is all AI-generated content slop?
No. AI can produce useful, substantive content when prompted well and edited thoughtfully. Slop is about quality patterns, not origin. Humans write slop too — the issue is low-effort, mass-produced content that prioritizes volume over value.
Where did the term 'slop' come from?
The term emerged on platforms like 4chan and Hacker News around 2022 for AI-generated images. Programmer Simon Willison championed it in a viral May 2024 blog post, arguing it should become the standard word like 'spam' did for unwanted email. By late 2025, three major language authorities had named it Word of the Year.
How is AI slop different from spam?
Spam is unwanted content sent to you (email, messages). Slop is low-quality content you encounter while actively seeking information (search results, articles, social media). Simon Willison proposed 'slom' for AI-generated spam, recognizing these categories overlap but remain distinct.
How much of the internet is AI slop?
A 2025 Graphite study found 52% of newly published articles online are AI-generated. Ahrefs found 74.2% of new web pages contain AI content. NewsGuard identified over 2,089 undisclosed AI-generated news websites. However, Google's algorithms filter most of it: only 14% of Google Search results are AI-written.
What is 'workslop'?
Workslop is AI-generated workplace content: generic emails, padded performance reviews, recruiter messages that match job postings suspiciously well, and marketing copy that's polished but forgettable. It creates 'productivity theater' — high output that masks downstream cleanup work.
What is slopsquatting?
A cybersecurity attack where hackers exploit AI code generators that hallucinate fake software package names. Attackers register these phantom packages with malicious code, then wait for developers to install them. Research found 43% of hallucinated names recur consistently, making them exploitable targets.
Can I use SlopDetector to check my own writing?
Absolutely. It's useful for identifying patterns in your writing that might feel generic or unsubstantiated. Think of it as feedback on quality signals, not a judgment on you as a writer.
Ready to Detect AI Slop?
Now that you know what AI slop is, try our free detector.