Per-sentence heatmap
See which sentences push the score up instead of guessing from a single percentage.
Paste text or drop a PDF, DOCX or TXT file and the AI detector scores how machine-written it reads — then shows the evidence: a heatmap of the sentences driving the score, the signals behind it and a concrete plan for the passages that read least like you. No black-box number.
Free account: 50 tokens every month · no card needed
See which sentences push the score up instead of guessing from a single percentage.
Several independent models are combined, with a false-positive guard that tempers borderline calls.
A free on-device estimate updates while you write; run the full check when the draft is ready.
Score many documents at once — useful for a course, a special issue or a set of chapters (paid plans).
Get specific suggestions for the flagged passages, and hand them to the AI Humanizer if you want help rewording them.
Keep the result with the manuscript's history or share it as a report (paid plans).
No detector can prove it. Scores are probabilities, and careful human writing can occasionally read machine-like. That is why this detector shows the sentences and signals behind the score — treat it as evidence to look at, not a verdict.
It looks at statistical and stylistic properties of the text rather than a fingerprint of one model, so it responds to text from the common chat assistants in general.
At least a few sentences; longer passages give steadier results. Very short text is hard for any detector to judge.
Every tool opens on the free plan: a free account gets 50 tokens every month. Longer documents, downloads and some advanced modes need a paid plan — the tool tells you when you reach one.
Yes — drop a PDF, DOCX or TXT file, or several at once in batch mode.
Open it in your browser — nothing to install.