Skip to content
ChatGPTDetect
English

AI detector for teachers

Teachers can check a student text for ChatGPT writing with a style-based detector such as ChatGPTDetect. They cannot use OpenAI's watermark detector, because OpenAI puts checking student work outside the scope of its program.

Why the watermark does not help in a classroom

OpenAI's textGrain watermark applies only to ChatGPT text in the EU, and its detector is restricted to approved researchers and expert organizations. Text from the rest of the world is not marked, and no mark does not mean a person wrote it. Details are on the ChatGPT watermark status page.

What a style-based check gives you

Paste the text and the checker reports whether it reads as AI-written, highlights the passages that carry the signal, and shows how close the style is to ChatGPT and other families. At least 100 words are needed. Measured on human-written text in seven languages, 0.5% was wrongly flagged as AI.

What to do with a result

Schools and larger volumes

The free check covers one text. For repeated checks, files and records you can keep, the paid tier of the same engine covers them. For the wider field, see the best ChatGPT detectors, and for the phrases that give AI text away, how to tell if ChatGPT wrote it.

Questions and answers

Can teachers use OpenAI's detector?

No. OpenAI's text detector is not public, and it lists academic-integrity checking of student work as outside the scope of its program.

How accurate is the detector on student writing?

Measured on human-written text in seven languages, 0.5% was wrongly flagged as AI, and 98.3% of fully AI-written texts were caught. The facts page explains what the figures measure.

Should a flag be treated as proof?

No. The result is writing-style similarity, a signal rather than proof of authorship. Use it as a reason for a conversation.

Is there a school or district option?

Yes. Fileverity quotes education pricing per institution, and the free check here works for a single text in the meantime.