How Turnitin's AI detection works

Turnitin's AI writing indicator splits a submission into overlapping segments and scores each one on how likely it is to have been produced by a large language model. It then reports an overall percentage and highlights the segments that drove it.

It only scores longer prose. Short submissions, lists, tables and code are skipped or produce no result. Low scores between 1% and 19% are shown with an asterisk instead of a number, because Turnitin says false positives are more likely in that range.

Can Turnitin detect ChatGPT after paraphrasing?

Sometimes. Turnitin added a separate check for text modified by paraphrasing tools, and in August 2025 it announced detection aimed specifically at AI bypasser and humanizer tools. Crude word-swapping tools are the easiest to catch, because they leave their own odd patterns.

Independent testing of these newer checks is limited. What is consistent across studies is that detectors are strongest on raw AI output and weakest on text a person has substantially rewritten.

AI detection result showing 72% AI detected for a paragraph about a bakery's exit strategy

Try it on your own text

Paste a paragraph and see your AI score before and after.

Check any paragraph for AI in one tap.

Download on the App Store

How accurate is Turnitin's AI checker?

Turnitin reports a low false-positive rate for documents with more than 20% AI writing. Independent researchers have found more mixed results, and accuracy drops on short texts, on writing by non-native English speakers and on highly formal prose.

Several universities have reacted by limiting how the score is used. Vanderbilt disabled the indicator in 2023, and others tell staff not to treat it as standalone evidence.

What this means for you

If you used AI in a way your course allows, make the final text yours. Rewrite in your own words, add your own examples, and check the result before you submit. TextHumanizer shows an AI score for any paragraph, which tells you which parts read as machine-written.

If you wrote the work yourself and still got flagged, keep your drafts and version history. That evidence matters far more than any percentage.

Questions

Can Turnitin detect Claude or Gemini?

Turnitin's model is trained to recognise text from large language models in general, not only ChatGPT. Output from Claude, Gemini and similar tools can be flagged too.

Can students see their Turnitin AI score?

Usually not. In most setups only instructors see the AI writing report, though some institutions share it with students.