AI text detector
A general-purpose checker for any written English: articles, emails, reports, product copy or essays. It reports a likelihood with an explicit uncertainty range, and lists every signal that contributed, so you can see whether the answer rests on something solid or on a coin-flip.
Longer samples give better answers. Below roughly 150 characters there is not enough signal for any method to work, so the tool declines rather than guessing. Even at length, treat the output as one input among several.
Drag a file here, or
JPEG, PNG, WebP, AVIF, PDF or .txt · up to 10 MB · you can also paste a screenshot
By uploading you agree to our Terms and Privacy Policy. Your file, IP address, approximate location and device details are stored.
What this checks
- A transformer classifier applied across the whole document in sections
- Sentence-length variance, which is consistently lower in generated prose
- Vocabulary diversity and repetition patterns
- Assistant-typical phrasing density
- Agreement between sections, as a check on mixed authorship
Frequently asked questions
How much text do I need?
At least 150 characters, and ideally several hundred words. Accuracy improves substantially with length because the statistical measures need enough sentences to be meaningful.
Does paraphrasing defeat it?
Generally yes. Running generated text through a rewriting tool, or editing it by hand, disrupts exactly the patterns detectors rely on. This is a fundamental limitation of the approach, not something a better model fixes.
Why show a range instead of one number?
Because a single number implies a precision the underlying methods do not have. The range reflects how much the available signals actually pin the answer down.