Check if text was written by AI
Every other detector hands you a percentage and no reasoning. This one shows which measurable properties produced the number, so you can judge whether it deserves any weight.
Read the number honestly
No detector can prove authorship, and anyone telling you otherwise is selling something. What these tools measure is how smooth and predictable a piece of writing is. Assistant output scores that way. So does careful, formulaic human writing, and so does writing by anyone working in a second language.
A high score is a question
It means the writing resembles machine prose on the axes we can measure. It is not a finding about what you did.
Short samples are noise
Under roughly 180 words, every detector on the market becomes unreliable. We say so on the result rather than hiding it.
Images can do better
Unlike text, images can carry signed provenance. Our image check reads facts, not guesses.
Flagged for something you wrote?
This is the most common reason people land here. There is a right order to do things in, and some of the evidence you need expires if you edit the file first.
- Your essay was flagged as AI and you wrote every wordA calm, practical walkthrough for students whose own writing got flagged by an AI detector: why it happens, what evidence actually helps, and how to respond.
- A professor has accused you of using AIHow to respond when an instructor accuses you of using AI: what to say, what evidence to bring, what your rights usually are, and the mistakes that make things worse.
- How to prove you wrote it yourselfThe evidence that actually demonstrates human authorship — version history, drafts, process artefacts — and how to set it up before you ever need it.
- What your Turnitin AI score actually meansWhat the Turnitin AI writing percentage measures, how it differs from the similarity score, what counts as high, and why it is not proof of anything on its own.
- Why AI detectors flag human writingThe mechanics behind AI detector false positives: perplexity, burstiness, threshold choice, and the specific groups of writers who get flagged most.
- What to actually say to your professorWording that works when responding to an AI accusation, plus what to say in the meeting and what to avoid, with a template you can adapt.
The detectors people actually face
What each tool measures, who it is sold to, what it gets wrong, and what its accuracy claim is actually based on.