Why we built this
AI detectors are now used to make decisions about grades, jobs, and reputations, yet most detector websites say little about how often they are wrong. We think people deserve straight answers: what a detector measures, how accurate it really is, and what to do when it gets things wrong.
We also think detection is the wrong tool for the most important question, “did this person write this?” That is why we built Prove Human, which records how writing is created and gives writers evidence they control.
How our detector works
- Engine. Text is scored by a commercial AI detection model from Sapling, which returns an overall score and sentence-level scores. We may change or combine engines over time and will update this page when we do.
- Scoring bands. Below 35% we show “Mostly human-like patterns,” from 35% to 70% “Mixed signals,” and above 70% “Strong AI-like patterns.” Every result explains what it can and can’t mean.
- Limits. We require at least 80 words because shorter samples are unreliable, and check up to 1,500 words at a time.
- No storage. AblePro does not store the text you check.
- What we will never do. Claim our detector proves anything, or present a score without its limits.
How our articles are written and reviewed
Every article lists its sources and the date it was last reviewed. We rely on published research and statements from detector makers and institutions, and we recheck facts at least quarterly because this field changes fast. If you find something out of date, please tell us.
No ads, no affiliate links
We don’t run ads or take payment to recommend detectors or “humanizer” tools. The site is supported by Prove Human subscriptions.