Direct answer
Turnitin checks for both. A single submission generates two separate reports: a similarity (plagiarism) report that matches your text against its comparison database, and an AI writing report that estimates how much of your qualifying prose looks machine-generated [1]. The two run independently, so a paper can score low on similarity and still carry a visible AI indicator [1].
Does Turnitin detect AI writing, or does it only check plagiarism?
Turnitin does both, and it does them as two separate analyses rather than one blended score. When a paper is submitted through an institutional integration, the similarity report and the AI writing report are opened from the same submission view, which is why instructors can reference either one without running a second check [2]. The similarity side answers "does this text overlap with existing sources?" while the AI side answers "does this text read as machine-generated?" — different questions, different outputs [2].
The AI writing report highlights individual sentences the model considers likely AI-generated and then reports a document-level percentage of qualifying text [2]. That means the AI indicator is not a plagiarism signal at all: text you wrote entirely yourself can be flagged, and text you lifted from a source can pass the AI check while failing similarity [2].
Turnitin also gates the display of the AI figure. The indicator is only shown when at least 20% of the analyzed text is flagged, and the company consistently describes the result as an indicator rather than proof of misconduct [2]. So "Turnitin checks for AI" is accurate, but it checks in a thresholded, probabilistic way — not as a yes/no verdict.
What does the Turnitin AI writing report show, and how is the AI percentage calculated?
The AI percentage is the share of your document's qualifying prose segments that the model classifies as AI-generated — it is not a count of words copied from any source [3]. Only longer-form prose is analyzed, so headings, bullet lists, short phrases and citation strings are excluded from the calculation [3]. That exclusion matters: a paper can look heavily structured and still produce a modest AI figure simply because much of its content never entered the analysis.
Because the calculation is confidence-based, low-confidence results are displayed as a symbol rather than an exact number [3]. This is the same mechanism behind the asterisk-style display you may have seen on a report — it signals that detection was below the confidence threshold, not that the document is clean [3].
Practically, the report gives you two things: a document-level percentage and sentence-level highlights showing which passages drove that percentage [3]. Read together, they tell you where the flagged prose sits, which is far more useful than the headline number alone when you are deciding what to revise.
Can I see my own Turnitin AI and similarity report before I submit?
Usually not through your institution. The instructor-facing report is the authoritative version, and students typically only see a report if their instructor has explicitly enabled the student view [4]. If that setting is off, you can submit a paper, receive a grade, and never see the AI or similarity breakdown behind it.
Turnitin's own guidance frames the AI indicator as a starting point for a conversation about authorship rather than a verdict, which is precisely why the company does not position it as a student self-check tool [4]. The report is designed for an instructor reading a class set of submissions, not for a student auditing a single draft under deadline pressure.
That gap is the practical problem. If you want certainty about your own draft before it enters the institutional system, you have to verify it separately — because once you submit, the report that matters is the one your instructor controls [4].
Knowing that Turnitin runs an AI check alongside the similarity check is only useful if you can see the result on your own draft first. turnitin0 lets you upload your.docx,.pdf or.txt and download the same two reports — the AI detection report and the similarity report — that your professor will see, in under 15 minutes in 98% of cases, without adding your file to Turnitin's student paper database.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
FAQ
Does Turnitin check for AI if my instructor only asks for a plagiarism check?
The AI writing report is generated from the same submission as the similarity report, so the capability is present regardless of what the assignment brief says [1]. Whether you see the AI indicator depends on your instructor's settings and on whether the document crosses the confidence threshold [2].
Can Turnitin flag fully human-written text as AI?
Yes. The AI indicator is a probabilistic estimate, not proof, and Turnitin itself describes it as a signal for discussion rather than a determination of misconduct [2]. Because only longer-form prose is analyzed and the result is thresholded, false positives are a known limitation rather than a rare edge case [3].
Does a low similarity score mean my AI score will also be low?
No. The two checks measure different things: similarity compares your text against a comparison database, while the AI report estimates how machine-generated your prose reads [1][3]. A draft can show a very low similarity percentage and still carry a visible AI indicator.
Why does my report show a symbol instead of an AI percentage?
That display appears when detection falls below Turnitin's confidence threshold, so the system declines to publish an exact figure [3]. It is a low-confidence signal, not a guarantee that the text is human-written.
How can I check my own AI and similarity results before submitting?
Institutional reports are instructor-controlled and the student view is optional, so the reliable route is to verify your draft with a separate pre-submission check that returns both the AI detection report and the similarity report [4]. turnitin0 does this without adding your file to Turnitin's student paper database.