Free Tool · AI Writing
AI Slop Checker
Paste a paragraph and see how many of the 742 known AI tells are hiding in it. This runs on the same 742-entry corpus AIStoryHub uses to self-check its own social media drafts before they go out, not a separate demo. The score you get here is the real one.
Scored entirely in your browser against a static word list. Nothing you paste is sent to a server, logged, or run through an AI model of any kind.
Questions
How it actually works.
Does the AI Slop Checker send my writing anywhere?
No. It runs entirely in your browser against a static, downloadable word list. Nothing you paste is uploaded to a server, logged, or processed by an AI model of any kind, unlike tools that route your text through a third-party API to check it.
Does it use ChatGPT or another AI model to check my writing?
No. The checker matches your text against the 742-entry Corpus of AI Clichés using pattern matching, not a language model. That's deliberate: a fixed, published word list gives a consistent score every time, instead of a judgment call that can vary between requests.
Is the AI Slop Checker free?
Yes. No account, no login, no email required. Paste your text and get a score immediately.
How is the AI Slop score calculated?
Each match is weighted by how strong a tell it is (0 to 100) and how many of them show up per 1,000 words, so one incidental word in an otherwise clean chapter doesn't sink the score the way a cluster of strong tells does. Terms the corpus marks as only meaningful on repetition need to appear at least twice before they count.
What counts as an AI tell?
Words and phrases AI models reach for disproportionately often, like "delve" or "tapestry", plus common sentence-level patterns like "It's not just X, it's Y." Each entry in the corpus is tagged red, orange, or yellow for how strong a signal it is, built from published frequency studies and direct testing against current models.
Does a high AI Slop score mean my writing is bad?
Not necessarily. A high score means specific words and patterns are statistically overrepresented in AI-generated text, not that the writing itself is weak. The real risk is reader perception: once a few of these tells get noticed, readers start reading the rest of the piece more skeptically, even the parts that are genuinely good. The score is a proxy for that risk, not a grade on craft.
Can I see the full list of AI clichés it checks against?
Yes. The entire 742-entry corpus is public, searchable, and downloadable as JSON at aistoryhub.co/corpus.