Check your text for AI slop before you publish
Paste a blog article, LinkedIn post or Reddit post. We mark the generic, template-like passages, show what is missing for readers, and suggest specific edits. It checks writing quality, not who or what wrote it.
This tool flags generic writing patterns and gaps in evidence. It does not detect whether AI wrote the text, and it is not a Google score or a ranking prediction. The free check saves your email, score and finding titles. Requesting a full text review also shares and saves your submitted text and results with LightSite for review and follow-up.
Checking a live website instead? Use the AI Slop Detector for websites.
How the AI slop text checker works
What we mean by AI slop
AI slop is generic, low-effort writing: text that sounds finished but says little a reader can use. People and tools both produce it. A high score means the text reads templated, not that a machine wrote it.
Research on slop treats it as a quality problem, measured through relevance, density, specificity and coherence, rather than an authorship label (Shaib et al., 2025).
Two layers of checks
Pattern checks run the same way every time and point to exact passages: stock hooks, "it's not X, it's Y" framing, clusters of hype words, filler transitions, flat sentence rhythm, dash density, repeated bullet frames, three-item lists, slogan endings, engagement bait, unsupported claims and self-promotion.
An editorial review then judges the writing itself: rhythm and sentence-length variety, information density, stock vocabulary, template rhetoric and voice. It never fact-checks, never asks for sources or images, and every finding must quote your actual text or we drop it.
Each platform is judged on its own terms
Blog articles: does it say something specific early, does the rhythm vary, is the vocabulary the author's own. LinkedIn posts: a real observation and a genuine tone instead of template hooks and engagement bait. Reddit posts: reads like a person talking to the community, not marketing copy pasted into a forum.
What never changes your score
Writing in the first person, typos, informal tone, short sentences, personal stories and length do not raise or lower the score. A sourced technical explainer is not marked down for lacking a personal anecdote. We never suggest adding typos or inventing stories to sound human.
How the score works
Each finding adds a capped amount for its category, so one noisy pattern cannot dominate. Severe bands need evidence across several categories, and short texts get a capped score with limited confidence. Lower is better: under 20 is specific and it shows, 20 to 44 has a few generic patterns, 45 to 69 is noticeably templated, 70 and above reads like a template with the name filled in.
How we word the feedback
Findings are written in a dry editor's voice: direct, specific, occasionally wry. The humour points at the writing, never at you, and never at the cost of accuracy. Fixes stay practical, and no finding claims a machine wrote your text.
Sources
- Google Search Central: Creating helpful, reliable, people-first content
- Wikipedia: Signs of AI writing (field guide)
- Shaib et al. (2025), Measuring AI "Slop" in Text
Frequently asked questions
Is this an AI detector?
No. It flags generic, template-like writing and missing evidence. Good human writing can pass and weak human writing can fail, whoever wrote it.
Is the AI slop text checker free?
Yes. Paste up to 40,000 characters, choose the platform and get results in about 20 seconds with no signup.
Do you store the text I paste?
The free check does not store your text. If you request a full text review, we save the submitted text, email and results for review and follow-up.
Will fixing these issues improve my Google rankings?
We cannot promise that. The score is an editing aid for writing quality, not a ranking prediction.
How is this different from the AI Slop Detector?
The AI Slop Detector checks a live website by URL. This checker reads a draft you paste, before it is published.
Related
Review and share your results
Get a full text review using the email form after your results. Share your email, submitted text and scan results with LightSite for review and follow-up.
Share by LinkedIn, X, email or copy. Where a public report is available, share its link; otherwise share the score and checker link.