Direct answer
Turnitin detects AI-generated content with a proprietary predictive model that analyses patterns of word choice and sentence structure across your whole submission, then reports the result as an overall AI writing percentage plus colour-coded highlighted segments [1]. You do not run the detection yourself: an instructor opens the AI Writing Report from the Similarity Report sidebar, and the score only appears for submissions longer than 300 words [1]. The percentage tells you how much of the text the model classifies as AI-generated, and anything below the 20% confidence threshold is displayed as an asterisk (*%) rather than a number [3].
How Does Turnitin Detect AI-Generated Content in a Submitted Paper?
Turnitin's AI writing detection is a predictive model, not a lookup against a database of known AI text. The system evaluates the statistical texture of your prose — how predictable each next word is, and how much sentence length and rhythm vary — and converts those signals into a probability that a passage was machine-generated [2]. This matters for how you read the result: the report is an inference about style, not a record of what tool you used.
Because the method is probabilistic, the vendor is explicit that false positives can occur. Writing that is highly formulaic, heavily templated, or produced through repeated paraphrasing of the same source can resemble machine output closely enough to be flagged [2]. That is one reason the report is designed to be read alongside the highlighted passages rather than as a single number.
The model also has defined boundaries. Text in languages other than English, and submissions that are largely technical notation, mathematics, or code, are excluded from the AI score because the underlying signals are unreliable there [2]. A blank or missing AI score therefore does not automatically mean "clean" — it can mean "not scored."
- The report is opened from the Similarity Report, in the sidebar panel [1].
- Only submissions exceeding 300 words are processed for AI detection [1].
- The model distinguishes AI-generated text from AI-paraphrased text in its highlighting [1].
What Does the AI Writing Score Percentage and the Highlighted Text Actually Mean?
The headline percentage represents the share of your submission that the detector identifies as AI-generated, calculated against the total word count of the scored text [3]. It is a proportion, not a grade and not a verdict — a 34% score means roughly a third of the assessed words carried AI-like signals, not that a third of your ideas were unoriginal. Reading it that way prevents the most common misinterpretation students make when they first open a report.
The asterisk convention is the detail most often missed. When the AI score falls below Turnitin's 20% confidence threshold, the report displays *% instead of an exact figure, signalling a low-confidence result rather than a precise measurement [3]. Treat *% as "the model is not confident enough to commit to a number," which is materially different from a confirmed low score.
The highlighted text is where the report becomes actionable. Each flagged segment can be expanded to reveal the specific passages the model attributed to AI, so you can compare the highlight against your own drafting history [3]. If a highlighted sentence is one you wrote and revised by hand, that is exactly the kind of context an instructor needs — which is why the vendor frames the score as a starting point for conversation rather than proof of misconduct [4].
| What you see | What it means | What to do next |
|---|---|---|
*% |
Score below the 20% confidence threshold [3] | Note it as low-confidence, not zero risk |
| A percentage figure | Share of scored text classified as AI-generated [3] | Open the highlights and locate each flagged segment |
| Highlighted passages | Specific text the model attributed to AI [3] | Compare against your drafts, notes, and version history |
| No AI score at all | Text excluded (non-English, math/code) or under 300 words [1][2] | Confirm the submission was actually scored |
Can I See My Own Turnitin AI Detection Report Before I Submit?
In most institutions, no — students cannot self-submit to the institutional Turnitin account their university licenses. Access to the AI Writing Report sits with the instructor, and whether students ever see the AI score is a setting the institution controls [4]. This is the structural reason so many students reach the deadline without knowing what their report will say.
That gap is precisely why educators are advised to treat AI scores as a discussion prompt rather than a finding. Guidance aimed at teaching staff recommends interpreting the percentage in context — assignment type, drafting process, student history — instead of issuing a judgement on the number alone [4]. If you can articulate how a piece was drafted, you are in a much stronger position than someone defending a score with no evidence either way.
The practical consequence is that pre-submission visibility has to come from somewhere other than your course account. Students who want to know what the report will look like before the real deadline typically use an independent checking service that returns the same style of AI and similarity output — cover page, overall score, flagged segments, and similarity summary — on their own draft, then decide whether to revise or leave the text as it stands.
If you would rather walk into submission knowing the number instead of guessing at it, turnitin0 gives you that look ahead: upload your .docx, .pdf, or .txt draft and receive a Turnitin AI detection report and a similarity report together, in the same format your instructor sees, usually in under 15 minutes. Files are checked without being added to Turnitin's student paper database, and you can delete them from your account afterwards.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
FAQ
Does Turnitin detect AI content by matching it against a database?
No. It uses a predictive model that assesses patterns of word choice and sentence structure rather than comparing your text to a stored corpus of AI writing [2]. That is why the same passage can score differently after substantial revision.
Why does my report show an asterisk instead of a percentage?
An asterisk (*%) is displayed when the AI score falls below Turnitin's 20% confidence threshold, meaning the result is a low-confidence signal rather than an exact measurement [3]. It should not be read as a confirmed zero.
Can Turnitin flag text I wrote myself?
Yes, false positives are possible — particularly with formulaic, heavily templated, or repeatedly paraphrased writing [2]. Keeping drafts and version history is the most reliable way to demonstrate authorship if a passage is questioned.
Is a high AI score proof that a student used AI?
No. Turnitin's own guidance frames the score as a prompt for discussion rather than a determination, and instructors are advised to weigh it against assignment context and drafting history [4].
Why is there no AI score on my submission?
AI detection only runs on submissions longer than 300 words, and text in other languages or dominated by mathematics and code is excluded from scoring [1][2]. A missing score usually means the text was not assessed, not that it passed.