Find the AI tells in your text. Free, instantly, nothing uploaded.
Paste a draft and every mechanical sign of machine writing lights up as you type: the stock words, the staged opener, the dash doing the work of a comma, the “research shows” with no research behind it. Each mark links to why it counts. The check runs in your browser, so there is no limit and nothing leaves your machine.
A checklist, not a verdict.
AI detectors guess whether a machine wrote something and give you a percentage. This does something more useful: it shows you the exact phrases that make text read as machine-written, using the patterns Wikipedia’s editors documented while cleaning up AI-generated articles. You see what to change, not just that something is off.
Strong tells count on one sighting. Weak ones, like a single dash, only count when they have company, because a careful writer might use any one of them on purpose. The checker applies that rule the same way the rubric does.
- Is it really free and unlimited?
- Yes. The check runs in your browser using pattern matching, so it costs nothing to run and there is no word limit, no signup, and nothing is sent to a server.
- Is this an AI detector?
- No. A detector guesses whether a machine wrote the text. This shows you the specific phrases and habits that make text read as machine-written, so you can fix them. It is a checklist, not a verdict.
- What does it look for?
- The mechanical signs from Wikipedia's Signs of AI writing: not-X-but-Y contrasts, staged openers like 'let's dive in', stock words like 'delve' and 'testament', borrowed authority like 'research shows', dashes everywhere, forced triads, chatbot leftovers, and more.
- Why did it miss something obvious?
- Some patterns need reading rather than matching, such as a closing line that only repeats the paragraph or a claim inflated without any of the usual phrases. Those are what the editor is for.
- Will a clean result pass Turnitin or GPTZero?
- This tool makes no claims about detectors and is not built to beat them. It is built to make writing read like a person wrote it. Detector outcomes are a different question.