Direct answer
An AI text detector (in Finnish, AI-tekstintunnistin) is a tool that analyses a piece of writing and estimates how much of it was likely produced by a generative AI model such as ChatGPT, Claude or Gemini. Turnitin's AI writing detector does this by scoring the share of qualifying prose in a document and reporting it as a percentage inside the AI Writing Report, alongside the separate similarity (plagiarism) report [1]. The number is an indicator for instructors to interpret, not a verdict that a student cheated [2].
What does a Turnitin AI text detector actually detect, and what does its AI score mean?
Turnitin's detector works at the level of text segments, not whole documents. The model reviews the submitted file, identifies prose that qualifies for analysis, and then reports what proportion of that qualifying text carries AI-writing characteristics [2]. This is why two essays of identical length can receive very different percentages: the denominator is qualifying text, not total word count.
The percentage is best read as a share, not a probability. A report showing 42% does not mean there is a 42% chance the student used AI; it means roughly 42% of the analysed text was flagged by the model [2]. Turnitin itself frames the indicator as one signal among several, to be weighed by an academic who knows the assignment, the student and the drafting history.
What the detector does not do matters just as much. It does not identify which model wrote the text, it does not prove intent, and it does not replace a conversation between instructor and student. References, bibliographies and heavily quoted passages are typically excluded from the qualifying word count, so a literature-heavy essay is judged mainly on the student's own sentences [2]. Understanding that boundary is the difference between reading a report calmly and panicking at a number.
For Finnish-speaking students writing in English, one practical point follows: the detector is language-sensitive in the sense that the text must be in English to be analysed reliably by the institutional tool most universities use. If your assignment is in Finnish, the AI report may be less informative, and your institution's policy — not the tool — decides what counts as misconduct [1].
Why does Turnitin sometimes show an asterisk (*%) instead of an exact AI percentage?
The asterisk is Turnitin's way of saying "low confidence, small signal." When the detected share of AI-written text falls below the model's confidence threshold — currently under 20% — the report displays an asterisk in place of a precise figure [3]. It is not a zero, and it is not a hidden high score; it is a deliberately vague display for a result the model cannot state with confidence.
Students often misread this in both directions. Some assume an asterisk means "clean," and stop revising prose that genuinely reads as machine-generated in places. Others assume the asterisk conceals a real percentage that the instructor can see in full. Neither reading is safe: the asterisk signals a low-confidence, low-share result, and the honest response is to look at the highlighted passages in the report rather than at the headline number [3].
Institutions also differ in how they treat asterisk results. Some departments ignore them entirely; others ask staff to review the flagged segments manually before raising anything with a student [3]. Because the interpretation layer sits with your university, the most useful thing you can do is know your own programme's policy and, where a report exists, read the segment-level detail rather than the summary badge.
Can I check my own draft with a Turnitin AI detector before I submit it?
Yes — but the route matters, because a normal institutional submission is not a neutral test. When you upload work through your university's Turnitin assignment, the file is typically stored in the paper repository and the report becomes visible to your instructor [4]. That is fine for a final submission; it is a poor way to experiment with a draft you may still be rewriting.
The safer sequence is to understand your institution's rules first. Many universities permit draft checks, some explicitly discourage them, and a few treat repeated pre-submission uploads as unusual behaviour worth a conversation [4]. If you are unsure, asking your tutor directly is faster and lower-risk than guessing — and it is the kind of transparency that protects you if a report later looks ambiguous.
Once the rules are clear, a non-repository pre-check gives you the same information without the side effects: you see the AI percentage and the similarity matches on your own draft, revise the flagged passages, and submit only when the report reflects what you actually intend to hand in [4]. That is the practical value of a detector — not a score to chase, but feedback you can act on while you still have time to act.
Reading a Turnitin report for the first time is easier when someone shows you what the real thing looks like — the cover page, the AI score band, the flagged segments and the similarity summary side by side. turnitin0 does exactly that: you upload your draft (English, 300–30,000 words,.docx/.pdf/.txt) and receive two downloadable PDFs in one order — an AI detection report and a similarity report, matching what instructors see in their LMS — usually in under 15 minutes, without your file entering Turnitin's student paper database. Over 100,000 reports delivered and 20,000+ students in the US, UK, Canada, Australia, New Zealand and Ireland have used it to see their own numbers before the deadline does it for them.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
Get Real Turnitin AI & [Similarity Report](https://www.turnitin0.com/guides/us/checkers)
FAQ
Is an AI text detector the same as a plagiarism checker?
No. A plagiarism or similarity checker compares your text against a database of sources and reports matching passages, while an AI detector estimates whether the wording carries machine-generated characteristics [1]. Turnitin produces both reports separately, and a document can score low on similarity and high on AI detection at the same time.
Does an AI detection percentage prove a student used ChatGPT?
No. The percentage describes the share of qualifying text the model flagged, not the probability of misconduct [2]. Turnitin positions the indicator as evidence to be interpreted by an instructor who knows the context of the assignment.
What does the asterisk (*%) mean on a Turnitin AI report?
It appears when the detected AI share falls below the model's confidence threshold, currently under 20%, so the exact figure is withheld [3]. Treat it as a low-confidence signal and read the flagged segments rather than assuming a clean result.
Can I get flagged for checking my own draft?
That depends on your institution's policy, since institutional submissions are usually stored in the repository and visible to staff [4]. Ask your tutor before uploading drafts repeatedly, or use a non-repository pre-check that does not add your file to the student paper database.
Does the detector work on Finnish-language assignments?
Turnitin's AI writing detection is built around English-language prose, and the qualifying-text logic assumes English sentence structure [1]. If your assignment is in Finnish, the AI percentage is less reliable, and your department's academic integrity policy — not the tool — determines what counts as misconduct.