About this detector
This tool runs client-side, uses simple linguistic heuristics (sentence length, lexical diversity, punctuation & repetitiveness) and gives a heuristic score. It does not call any external API and cannot reliably prove authorship — treat results as guidance.
How to interpret the score
- 0–30: text reads like natural human writing (varied vocabulary, punctuation usage, sentence length variation)
- 30–60: mixed signals — parts may be formulaic or AI-assisted
- 60–100: text exhibits patterns commonly produced by current large language models (low diversity, uniform sentence length, repetitive n-grams)