Direct answer
Direct Answer - Turnitin detects two main things: text that matches existing sources (its similarity/plagiarism check) and text that looks AI-generated (its AI writing detection). The similarity check compares your submission against student papers, web pages, and academic periodicals, while the AI detector analyzes sentence-level writing patterns and reports the share of the document it flags as AI-written [1]. Both signals appear on the same instructor report, but they are calculated separately — so a draft can score low on similarity and still carry a high AI flag [3].
What does Turnitin's similarity report actually detect and match?
The similarity side of Turnitin matches your submitted text against three repositories: previously submitted student papers, indexed internet content, and subscription academic periodicals and journals [2]. Anything that overlaps with those collections is highlighted in the report, with a numbered citation pointing to the source it matched.
Critically, the similarity percentage is not a plagiarism verdict. Turnitin surfaces matching strings and leaves interpretation to the instructor, who can exclude quoted material, the bibliography, and small matches from the count [2]. That means properly cited quotations can still raise your percentage without implying misconduct.
The report also breaks the score down source by source, so an instructor sees exactly where each match came from rather than a single opaque number [1]. Common sources of legitimate matches include standard phrasing in a field, your own earlier submissions, and reference lists that were not excluded.
How does Turnitin detect AI-generated writing, and what does the AI score mean?
Turnitin's AI writing detection is a separate model from the similarity check. It analyzes the document at sentence level, looking at patterns associated with large language models, and then reports the percentage of the text it classifies as AI-generated [3]. This is why a fully original essay can still carry an AI flag.
The AI score has an important display quirk: when detection confidence falls below Turnitin's 20% threshold, the report shows *% instead of an exact figure [3]. A *% result is a low-confidence signal, not a clean bill of health, and instructors are trained to read it that way.
Because the two systems are independent, you can see any combination of results — low similarity with high AI detection, high similarity with no AI flag, or both low [3]. Students who only check for plagiarism therefore often miss the AI dimension entirely.
How can I see what Turnitin detects on my own draft before I submit it?
In most institutions, students cannot run an official Turnitin check themselves, because access is controlled through the instructor's LMS course [4]. That leaves pre-submission checking as the practical way to see what the report will show before the deadline.
Pre-submission checking is a widespread student habit precisely because fixing flagged passages early is far easier than appealing a flagged submission after the fact [4]. Once a paper is in the institutional system, the report is already visible to your instructor and the window to revise has usually closed.
The most useful pre-submission check reproduces the same two outputs an instructor sees: a similarity report with matched sources and an AI detection report with the flagged percentage [4]. Reviewing both together tells you whether the issue is unintended matching text, AI-style phrasing, or both — and each problem calls for a different fix.
Turnitin0 is built for exactly this pre-submission moment: it returns the same Turnitin AI and similarity reports your instructor will see, so you know what Turnitin detects on your draft while you can still change it.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
FAQ
Does Turnitin detect paraphrased text?
Yes — the similarity check matches overlapping strings, so close paraphrase can still register as a match against the original source [2]. Rewriting in your own words reduces the match length, but the safest approach is to cite the source regardless.
Is the AI score the same as the similarity score?
No. They are two independent systems: similarity measures overlap with existing sources, while AI detection measures sentence-level patterns associated with language models [3]. A paper can be flagged on one and not the other.
What does a *% AI score mean?
It means Turnitin's confidence in the AI signal fell below its 20% threshold, so the report shows an asterisk instead of a number [3]. Treat it as a low-confidence indicator rather than proof the text is human-written.
Can I check my own draft before submitting?
Usually not through your institution, since Turnitin access runs through the instructor's LMS course [4]. Independent pre-submission services exist that return the same style of AI and similarity reports so you can review them first.
Does Turnitin detect text copied from my own previous papers?
Yes — previously submitted student papers are part of the repository Turnitin matches against [1]. Reusing your own earlier work can therefore appear as a match unless it is properly quoted or the instructor excludes it.