Short answer: the Turnitin AI score is the share of the qualifying prose in your file (sentences in normal paragraphs) that Turnitin's model thinks was likely written by AI, or written by AI and then reworded with a paraphrasing tool (Turnitin: Using the AI Writing Report). To see how a detector reads your work before you submit, try a free AI detector. The Turnitin score is not the share of your whole document, it is separate from the similarity (plagiarism) score, and Turnitin says it should not be the sole basis for action against a student.
This guide explains what the number means, how it is calculated, why some reports show an asterisk instead of a number, and what to do if your work is flagged. Everything here comes from Turnitin's own published guides.
What the Turnitin AI percentage means
Turnitin's guide Using the AI Writing Report describes the overall percentage as the amount of qualifying text that its model determines "could be generated by AI", including AI text that may have been modified with an AI paraphraser or "bypasser" tool. Flagged text is highlighted in blue in the report.
Three things students often get wrong:
- It is not a plagiarism score. Turnitin's AI detection FAQ says the similarity score and the AI percentage are "completely independent and do not influence each other". A 5% similarity score and a high AI score can appear on the same paper, or the other way round.
- It is not a percentage of every word in the file. Only prose sentences in long-form writing count, so headings, lists and references are left out of the calculation.
- It is not a verdict. Turnitin states that it "does not make a determination of misconduct"; it provides data for the educator to decide under the institution's own policy (FAQ).
How Turnitin calculates the AI score
According to Turnitin's FAQ on its AI writing detection, the process works like this:
- Non-sentences are removed. The model drops text it predicts is not real sentences, such as lists, headers, references, data tables, formulas and code. These do not count towards the percentage at all.
- Each sentence gets a score between 0 and 1, judged in the context of the text around it (Turnitin FAQ).
- Sentences above a set threshold are labelled AI; the rest are labelled human.
- The document score is the percentage of evaluated sentences labelled AI (Turnitin FAQ).
This is why the percentage and the highlighted text sometimes do not seem to match. A document that mixes paragraphs with tables, bullet points or code can show a percentage that looks out of proportion to the highlights, as Turnitin explains in Using the AI Writing Report.
Which files get an AI score
A file only gets an AI percentage if it meets Turnitin's file requirements: at least 300 words of prose in long-form writing, no more than 30,000 words, under 100 MB, a .docx, .pdf, .txt or .rtf file, and a supported language (English, Spanish, Japanese, Arabic, German, Dutch or Turkish). Hindi and other Indian languages are not on that list. For PDFs specifically, see can Turnitin detect AI in a PDF.
Why your Turnitin AI score shows an asterisk (*%)
If you see *% instead of a number, the model found some likely AI text, but less than 20% of the qualifying text. Turnitin does not show a number or highlights in the 1-19% range because its testing found "a higher incidence of false positives" there (Using the AI Writing Report; Turnitin help center).
Reports generated before 8 July 2024 may still show a number below 20%, according to the same Turnitin guide. In plain terms, an asterisk means "low and less reliable", not "AI detected".
| What you see | What it means | What to do |
|---|---|---|
| 0% | No qualifying text was identified as likely AI (Turnitin). | Nothing. Keep your drafts anyway. |
| *% | Some text scored as AI, but below 20%, a range Turnitin calls less reliable. | Usually nothing; no number or highlights are shown. |
| 20-100% | That share of qualifying prose was flagged as likely AI or AI-paraphrased (Turnitin). | Read the highlights if you have access, gather your drafts, talk to your instructor. |
| Grey, no number | Not processed, often because the file did not meet the requirements. | Check word count, file type and language. |
Source for the states above: Turnitin, Using the AI Writing Report and How to access the AI Writing Report.
What does a Turnitin AI score of twenty percent mean?
Twenty percent is the lowest number Turnitin will display (Turnitin). It means the model labelled about one in five of your evaluated sentences as likely AI. Turnitin itself does not publish a "safe" or "fail" percentage; what counts as a concern is up to your instructor and your university's academic integrity policy.
Because the score is calculated per sentence, a 20% result on a long essay may come down to a few paragraphs. Ask which passages were highlighted, then look at your drafts for those sections.
Can Turnitin be wrong? False positives
Yes. Turnitin says its model "may not always be accurate" and can misidentify human-written, AI-generated and AI-paraphrased text, which is why it "should not be used as the sole basis for adverse actions against a student" (Using the AI Writing Report).
Turnitin's FAQ gives more detail:
- Its target false positive rate is under 1% for documents with over 20% AI writing, which it describes as possibly flagging one in every 100 fully human-written documents (Turnitin FAQ).
- Text that can trigger false positives includes writing with little structural variation, text that repeats itself, and text paraphrased without developing new ideas.
- Short documents of a few hundred words tend to get "all or nothing" predictions, so a mix of your own writing and AI-assisted text could be flagged as entirely AI.
- Grammar checkers: in Turnitin's tests, spelling and grammar fixes from tools such as Grammarly were in most cases not flagged, but text from generative features (drafting, paraphrasing, summarising) likely will be (Turnitin FAQ).
Turnitin also notes, in its model release notes, that it saw more false positives in generic opening and closing sentences and changed its detection logic to reduce them. For a broader look at why human writing gets flagged, read AI detector false positives.
Can students see their Turnitin AI score?
It depends on your instructor. Turnitin's FAQ says the AI indicator and report are visible to instructors and administrators, who can download the report as a PDF and share it (Turnitin FAQ). Its newer guide adds that if an instructor makes the AI Writing Report available, students can open it from their Similarity Report (Using the AI Writing Report). If you cannot see it, ask.
Turnitin and DrillBit in Indian universities
In India you may meet two different checkers. Many universities run theses and dissertations through DrillBit, which INFLIBNET's ShodhShuddhi programme has offered to higher education institutions since October 2023 (ShodhShuddhi), while Turnitin is used where an institution buys its own licence. They produce different reports, so a number from one tool says nothing about the other. Our comparison of Turnitin vs DrillBit vs iThenticate explains who uses which.
What to do if Turnitin flags your work
- Stay calm and ask what was flagged. Request the AI report or at least the highlighted passages. Turnitin's guidance for instructors calls the score a single data point, not a definitive answer (Turnitin).
- Collect your evidence. Version history in Google Docs or Word, earlier drafts, notes, outlines, research files and browser history for your sources.
- Talk to your instructor. Offer to explain your argument, your sources and how you wrote the piece.
- Be honest about tools you used. If you used a grammar checker, a translator or an AI tool for brainstorming, say so, and check what your course allows.
- Follow your university's process. If it becomes a formal case, ask for the policy in writing and keep copies of everything you submit.
Checking your work before you submit
You cannot run Turnitin's AI check yourself, and no other tool reproduces its score. A self-check is still useful as a second opinion: ProofAI's free AI detector shows an AI probability and the passages that drove it, with free checks every day and no login needed, and accepts PDF and Word files too. If a passage you wrote yourself reads as generic, the fix is to make it more yours: your examples, your reasoning, your sources. For a detailed comparison of how ProofAI and Turnitin differ, see our Turnitin alternative for India page. ProofAI, like every detector, can be wrong, so treat its result as a signal too.
FAQ
What is a good Turnitin AI score?
Turnitin does not define a pass mark. Below 20% shows as an asterisk because Turnitin considers that range less reliable (Turnitin); above that, your institution's policy and your instructor decide.
What does the asterisk (*%) mean on Turnitin?
The model found some likely AI text, but under 20% of the qualifying text, so Turnitin shows no number or highlights to avoid false positives (Turnitin help center).
Is the Turnitin AI score the same as the similarity score?
No. Turnitin says the two are completely independent (FAQ). Similarity measures matching text; the AI score estimates likely AI-written sentences.
Why did my AI score come out higher than the highlighted text suggests?
Only prose sentences in long-form writing are counted. Lists, tables, headings and references are removed first, so the percentage is of the remaining sentences, not of the whole file.
Will my Turnitin AI score change if I resubmit?
It can. Turnitin says the percentage for a submitted document changes only if it is processed again, and its model is updated over time (FAQ).
Conclusion
A Turnitin AI score is an estimate of how much of your qualifying prose looks AI-written, not a plagiarism score and not proof of misconduct. Scores under 20% are hidden behind an asterisk, and Turnitin itself says the result needs human judgment. Keep your drafts, know what your course allows, and if you are flagged, ask for the details and talk to your instructor.
Originally published on ProofAI.
Top comments (0)