Direct answer
A single scan can produce both signals: a similarity (plagiarism) report that highlights text matching existing sources, and an AI writing report that estimates how much of your prose looks machine-generated. Turnitin layers its AI indicator on top of the same submission that generates the similarity score, so one upload returns two separate reports rather than one blended number [1]. The two metrics answer different questions and are meant to be read together, not substituted for each other.
What does a combined plagiarism and AI scan actually check?
The similarity side compares your document against three source pools: the student paper repository, current and archived web pages, and a large index of publications [2]. Every overlapping string is highlighted and attributed, so you can see not just how much matched but exactly which sentence matched and where it came from. The percentage at the top is a sum of matched text — it is not a plagiarism verdict, and correctly quoted or cited material still counts toward it [2]. That is why a paper with a heavy literature review can legitimately show a double-digit similarity score.
The AI side is a different measurement entirely. It looks for statistical patterns in word choice and sentence construction that are characteristic of generated prose, and it reports the share of qualifying text that carries those patterns [1]. Turnitin's own guidance is explicit that this is an indicator, not proof: instructors are told to interpret it alongside the similarity report and their knowledge of how the student writes [1].
Practically, that means a combined scan gives you two independent lenses on one draft. The similarity report tells you whether your sourcing and paraphrasing are doing their job; the AI report tells you whether your own sentences read as distinctively human. A draft can pass one and fail the other, and that is normal rather than contradictory [2].
Why do Turnitin plagiarism and AI scores sometimes disagree with each other?
Because they are not measuring the same thing. Similarity measures overlap with existing text; the AI indicator measures stylistic regularity [3]. A passage you wrote entirely yourself can be 0% similar and still trip the AI indicator if it is unusually uniform in rhythm and phrasing. Conversely, a paragraph dense with quoted definitions can push similarity high while the AI indicator stays low, because quoted material is handled differently in the two analyses [3].
Low-confidence signals explain a lot of the confusion students report. When a document is short, heavily edited, or written in a mix of human and machine drafting, the AI model may not reach its confidence threshold — Turnitin then displays an asterisk instead of a precise percentage [3]. Seeing "*%" is not a clean bill of health and not an accusation; it means the evidence was too thin for a firm number.
Heavy editing also moves the two metrics in opposite directions. Rewriting AI-assisted sentences by hand tends to lower the AI indicator while leaving similarity untouched, and adding properly cited sources raises similarity while leaving the AI indicator untouched [3]. If you only look at one number after a revision, you can easily conclude nothing changed when the other metric moved substantially.
Can you get both reports from one scan before you submit?
Sometimes, and it depends on your institution. Whether a student can self-check through the institutional Turnitin account is controlled by the school's settings, and many students never see a preview before the graded submission [4]. If your course has that option enabled, you can run a draft and read the same report types your instructor will read.
The catch is the repository. A standard institutional submission is added to the student paper database, which means your own draft becomes a future match source — a real problem if you plan to revise and resubmit [4]. Non-repository checking avoids that: the draft is scanned without being added to the database and the report is not shared onward, so you can iterate on the same assignment without poisoning your own similarity score later [4].
What you want from a pre-submission scan is therefore specific: both report types, generated from one upload, without the file entering the paper database. That combination is what lets you act on flags — rework the flagged passages, fix the paraphrasing, re-check — while there is still time to change the outcome [4].
Most students only discover which passages were flagged after the deadline has already passed, when the report arrives as feedback rather than as a warning. turnitin0 exists to move that moment earlier: you upload your draft, and you get the AI and similarity reports back while the assignment is still yours to edit.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
FAQ
Does a similarity score of 0% mean my paper is safe?
No. A 0% similarity score only means no text matched the sources Turnitin checks, which says nothing about the AI indicator [3]. The two metrics are independent, so you need to read both before concluding a draft is clean.
Can my own earlier submission be counted as plagiarism?
Yes, if the earlier draft entered the student paper database, it can match your later submission [4]. Non-repository checking avoids this by scanning the file without adding it to the database.
Why does my report show an asterisk instead of a percentage?
Turnitin displays "*%" when the AI detection result falls below its confidence threshold, which happens with short documents, mixed authorship, or heavy editing [3]. It signals low confidence, not a clean result.
Does one upload really produce both reports?
Yes. Turnitin's AI writing detection is integrated into the same submission workflow that generates the similarity report, so a single upload returns both [1].
Should I rewrite everything the AI report flags?
Not necessarily. The AI indicator is designed to be interpreted alongside the similarity report and your instructor's judgement rather than treated as a verdict [1]. Focus on passages that are both flagged and stylistically unlike your own writing.