AI slop detector
Paste any text to see the AI slop in it, highlighted phrase by phrase, with a slop score from 0 to 100. Free, unlimited, and it runs in your browser.
What it flags
| Sign | Examples | Why it reads as AI |
|---|---|---|
| Slop phrases | "In today's fast-paced world", "it's important to note", "plays a pivotal role" | 100+ phrases from our measured list; they announce a point instead of making it. |
| AI vocabulary | crucial, delve, fostering, showcase, underscore, multifaceted, leverage, quietly | Words that studies (and Wikipedia's guide) find LLMs overuse, including the GPT-6-era set. |
| Negative parallelism | "It's not X, it's Y", "not just X, but Y", "Y rather than X" | Corrects a claim nobody made. See "it's not X, it's Y". |
| Corrective disclaimers | "should be treated as directional rather than definitive", "does not establish" | New in mid-2026 chatbots; Wikipedia added it as a sign this year. |
| Copula avoidance | "serves as a", "stands as a", "marks a turning point" | Where a person writes "is". |
| Chatbot offers | "If you'd like, I can…", "I hope this helps" | Text meant for the chat window, pasted into the document. |
| Formatting, rhythm | **Bold labels**, em dashes, lists of exactly three | Half weight: humans use these too. |
Slop score, not authorship
This is a slop checker, not an AI detector. It doesn't claim to know who wrote something and it isn't for catching students or beating Turnitin (see our acceptable use policy). It answers a simpler question: which sentences will make a reader think "chatbot"? Each highlight has a tooltip saying why.
Signs shift with each model. In our November 2026 test, ChatGPT's default model used 0.75 em dashes per 1,000 words, but the GPT-5.5 to GPT-6.1 API models used 3.8 to 9.2 (human answers to the same questions: 0.21). So an em dash on its own tells you little, which is why the detector gives formatting half weight and leans on phrasing: contrasts, hedges, offers and stock vocabulary.
Found slop? Fix it
Edit the highlights by hand, or paste the text into the unslop tool: it removes the phrases, rewrites in your tone, keeps your facts and numbers, and shows the slop score before and after.
Free: 3 AI rewrites a day (up to 300 words) and unlimited quick clean. Need more? A $5 pack of 50 rewrites never expires, or go Pro at $12/month · $79/year. See plans.
FAQ
What is an AI slop detector?
A checker that flags the habits of low-effort AI text ("AI slop"): filler phrases, overused AI vocabulary, "it's not X, it's Y" contrasts, chatbot offers, bold labels and rule-of-three lists. It scores how much slop the text has, not who wrote it.
Is this an AI detector?
No. It does not guess whether a human or a model wrote the text, and it makes no promises about Turnitin, GPTZero or any other detector. Plenty of people write "crucial" or use em dashes. It shows you the specific phrases that make text read like a chatbot so you can edit them.
Is it free?
Yes. The checker is free, unlimited and runs entirely in your browser. Nothing you paste is uploaded. If you want the slop rewritten for you, the unslop tool gives 3 free AI rewrites a day, then $5 for 50 or Pro at $12/month.
How is the slop score calculated?
We count matches of 100+ known slop phrases plus pattern checks (negative parallelisms, corrective disclaimers, AI vocabulary from Wikipedia's guide and published studies, copula avoidance, chatbot offers, formatting), weight formatting and rule-of-three hits at half, and scale per 100 words to 0–100. 16–39 means some tells, 40+ means heavy slop.
Does it work on LinkedIn posts?
Yes. LinkedIn added a "seems like AI slop" report option in 2026, and LinkedIn posts are where "it's not X, it's Y", rule-of-three lists and stacked one-line paragraphs show up most. Paste the post before you publish.