Claude text detector

How Claude text gives itself away

Claude is a model from Anthropic. People often use it for longer texts: research write-ups, document summaries or edits to work in progress. The output tends to be polished and well structured, and that regularity is exactly what detectors measure: sentences of similar length, the same paragraph structure and predictable word choice.

Claude's style changes with the prompt and with the model version, so the check does not rely on a list of its favourite words. It looks at the properties of machine writing in general. Accuracy may change with new model versions, though.

What the check cannot tell

Which model wrote the text. The result does not say that the text came from Claude specifically. The differences between models are smaller than the differences caused by the prompt or by later edits, and an independent detector has no key to their watermarks.

Nor can it reliably recognise text that someone thoroughly rewrote, translated or mixed with their own after generating it. With text like that the detectors disagree and the result is borderline.

What to watch for in long texts

Claude can handle a long document in one go, so it may be that only one chapter was generated and the rest is the author's own. A score for the whole text dilutes a part like that. So read the marked sentences and look for the places where the style changes.

Claude can also write from a sample of the author's text. The result may then come out less machine-like than its origin would suggest, so a low score is just one thing to go on.

What to do with the result

A high score is not proof that the text came from Claude, and a low one is not proof of the opposite. No AI text detector is one hundred per cent accurate.

When you check someone else's text, take the result as a basis for questions: why the author chose this source, what the term in the second paragraph means, how they reached the conclusion. When you check your own, it shows you the passages worth rewriting in your own words.

Can the detector tell that a text is from Claude specifically?

No. The result says how strongly the text carries the marks of machine writing, not which model wrote it. Text from Claude may come out the same as text from ChatGPT.

Can I check a whole long text from Claude?

The demo without signing in checks a section of up to 3,000 characters. With a free account you can paste a text of up to 300,000 characters and see where the style changes.

How much text is needed?

At least 100 characters; a continuous paragraph or a whole chapter is more reliable. With just a few sentences the result is indicative only.

Does it work on text from other models too?

Yes. The check measures the properties of machine writing in general, not the signature of a particular model, so you can check text from ChatGPT and other models in the same way.