Does Turnitin detect AI?
Yes, when your school has the feature switched on. Turnitin estimates how much of a submission was likely written by AI and shows that number to your instructor.
Pre-check your essay for AIThe short answer
Turnitin has had AI writing detection since April 2023. It runs alongside the similarity (plagiarism) check, but it is a separate feature that each institution chooses to license and enable. Some have it on, and a number of universities have turned it off, citing concerns about false positives.
So whether your work is checked for AI depends on your school and sometimes on your course. Your syllabus or instructor is the only reliable source for that.
How Turnitin’s AI detection works
The similarity check compares your text with a database of sources. The AI check does something different: it looks at the writing itself. Turnitin’s model reads the submission in overlapping segments and estimates, for each, how likely it is to have come from an AI writing tool. Those estimates are combined into one percentage for the document.
As Turnitin has described it:
- The percentage is the share of qualifying text, meaning prose sentences in long-form writing, that the model judges likely to be AI-written. Lists, tables, code and short answers are not what it is built for.
- A submission needs a minimum amount of prose to be scored. Turnitin’s documentation has put this at around 300 words.
- The report highlights the passages behind the number, and later versions separate text that looks AI-generated from text that looks AI-generated and then paraphrased.
- It started with English and has added other languages over time.
Turnitin changes these details between releases. Your institution’s Turnitin guide has the current ones.
What the Turnitin AI percentage means
It is an estimate of how much of the text may have been written by AI. It is not a measure of how much you used AI, and it is not a finding of misconduct. A university help page puts it this way: a high score "is not definitive evidence of academic dishonesty" and "should be considered a starting point for dialogue with the student".
Two details matter when reading a score:
- Low scores are not shown as a number. Turnitin has said that results from 1% to 19% are less reliable, and its report marks them with an asterisk instead of a figure.
- Turnitin states a false positive rate of under 1% for documents where it finds more than 20% AI writing. That is the company’s own figure. It also means that across thousands of submissions, some fully human papers will be flagged.
Can Turnitin be wrong?
Yes, in both directions, and Turnitin says so. It has acknowledged that it misses a share of AI-written text in order to keep false positives low.
- False positives. Human writing that is formal, formulaic or very even in rhythm can be flagged. Research on AI detectors in general has found more false alarms on writing by non-native English speakers. Turnitin disputes that its own model is biased in this way.
- False negatives. AI text that has been heavily edited or rewritten is harder to detect.
- Institutions disagree about using it. Several universities have disabled the feature, among them Vanderbilt and Johns Hopkins, pointing to the risk of falsely accusing students. Others keep it on as one signal among several.
Can students see their Turnitin AI score?
Usually not. The AI report is shown to instructors. Whether students see anything depends on how the institution has set Turnitin up, and most do not show it. Turnitin does not sell AI checks to individual students.
That is why outside pre-checks exist. Be clear about what they are, though: every one of them, including EssayPrecheck, uses its own detector and gives its own number. None can show you the score your instructor will see. See Turnitin checker for students.
What to do if Turnitin flags your work
- Stay calm and ask to see the report. Ask which passages were highlighted and what the percentage was.
- Gather your process evidence. Version history, drafts, notes, sources and browser history show how you wrote the piece. This is far more persuasive than any other detector’s score.
- Point to the limits of the tool. Turnitin’s own guidance says the score should not be the sole basis for action. Many university policies say the same.
- Offer to talk it through. Being able to explain your argument, your sources and your choices is strong evidence that the work is yours.
- Use your school’s process. If it becomes formal, follow the academic integrity procedure and ask an adviser or student union for support.
This page was last checked on November 11, 2026. It describes another company’s product, which changes often.
Questions and answers
Does Turnitin detect ChatGPT?
Turnitin’s AI detection is designed to flag text from large language models such as ChatGPT. It works on unedited output best and can miss text that has been heavily rewritten.
What percentage of AI is acceptable on Turnitin?
There is no universal threshold. Each institution or instructor decides how to treat a score. Turnitin shows scores from 1% to 19% as an asterisk because they are less reliable.
Does Turnitin detect paraphrased AI text?
It tries to. Later versions of the report mark text that looks AI-generated and then paraphrased. Detection of rewritten text is less reliable than detection of unedited output.
Does Turnitin flag Grammarly?
Turnitin has said that basic spelling and grammar correction is not meant to trigger its AI detection. Tools that rewrite whole sentences or generate text are a different matter and can be flagged.
Is every Turnitin submission checked for AI?
No. AI detection is a separate feature that an institution must license and enable, and some have turned it off.
How accurate is Turnitin’s AI detector?
Turnitin states a false positive rate under 1% for documents with more than 20% AI writing, and has acknowledged that it misses some AI text. Independent studies have reported mixed results.