AI Slop Detector
Paste a draft. Get a slop score out of 100 and every AI tell it found, line by line.
AI slop is writing with a model's fingerprints on it: a vocabulary no person types, lists of three, throat-clearing openers, an ending that restates the piece. This check scans your draft against 24 documented patterns, scores it, and hands back an edit with the patterns removed. It does not guess who wrote the text.
Single · free · 1 a day — Council and Deep on any paid plan
This usually takes about 5 seconds.
Deep takes a minute or two: four pro models answer, two critics review, one mediator writes the final call. This page updates when it lands.
Every free tool here runs one model. Pro runs four.
A single model agrees with itself. Multi-model consensus puts four of them on the same draft and shows you where they disagree, which is where the errors live. That runs on Pro. Leave your email and we'll send you the walkthrough, no account needed.
One email, no spam. Unsubscribe anytime.
Done. Check your inbox.
The multi-model walkthrough is on its way.
You've used your free analysis for today. Sign up for TrueStandard to get unlimited access with 4 AI models instead of 1.
Or come back tomorrow for another free analysis.
Here for the edit, not the score? The AI Humanizer runs the same check
What counts as an AI tell?
AI writing has a fingerprint. Not a vibe, a fingerprint: specific words, specific sentence shapes, specific structural habits that show up far more often in generated text than in writing a person actually typed.
We scan your draft against 24 documented patterns
Banned vocabulary like delve, leverage, robust and tapestry. Dead phrases like it's worth noting. Structural tells like the rule of three, throat-clearing openers, and endings that restate the whole piece. You get a score out of 100 before anything is rewritten.
We check the prose mechanics too
Passive voice that hides who acted. Adverbs propping up weak verbs. Sentences carrying three ideas and two commas. Reading grade. These are separate from AI tells, so you get a separate score for them.
We edit, then re-run both checks
The editor gets the exact list of faults, not a vague instruction to sound human. Then the same scanner runs again on the result, so you see the before and after side by side rather than taking our word for it.
The pattern catalogue comes from Wikipedia's Signs of AI Writing guide, maintained by WikiProject AI Cleanup, combined with nine rules of prose mechanics. Both run as deterministic checks, so the same text always gets the same score.
Cleaning up how it reads is one problem. Knowing whether it's actually true is the other one.
Before and after
In today's world, the landscape of creator monetization is undergoing a transformative shift. It's worth noting that many creators are now leveraging multiple revenue streams to foster sustainable growth. Here's the thing. Most creators delve into paid tiers far too early. The reality is that a seamless upgrade path takes 6 to 12 months to build properly. And that matters.
Creator monetization is changing. Many creators now run several revenue streams at once to keep growing. Most of them open paid tiers too early, before the free list can carry a paid offer. Building an upgrade path people actually take runs about nine months.
Frequently Asked Questions
Low-effort AI-generated writing that reads fine at a glance and falls apart on inspection. On the page it shows up as a fingerprint: a small vocabulary the models overuse, sentence shapes they repeat, structural habits like restating the whole piece in the last paragraph. The detector scores that fingerprint, not the ideas.
Two numbers. AI tells counts documented generated-writing patterns per 100 words, from a catalogue based on Wikipedia's Signs of AI Writing guide. Prose tightness measures mechanics: passive voice, adverbs, overlong sentences, reading grade. Both checks are deterministic, so the same text always gets the same score.
No. An AI detector guesses whether a machine wrote the text, and the commercial ones are wrong often enough to flag human writing. This check does not care who wrote the draft. It lists the specific patterns that make writing read as generated and removes them, so you can judge the result yourself.
Both, in one run. The model gets the exact list of faults rather than a vague instruction to sound human, edits the draft, and the scanner runs again on the result so you see the before and after side by side. Your numbers, names and quotes are meant to survive untouched, and the check flags any number that went missing.
One draft a day, up to about 300 words, on one model. Signed-in accounts on a paid plan run it against their credits with no daily cap, across four models, and see where the four edits disagree.
Powered by TrueStandard
Multi-model AI verification for high-stakes decisions.