Direct answer
Turnitin works in two connected stages when an instructor enables both features: it compares your paper against a massive database of web pages, academic publications, and previously submitted student work to produce a Similarity Report, and it runs a separate AI writing detection model that estimates how much of the document was likely generated by an AI tool [1]. Both results appear in one report interface inside the instructor's learning management system, showing a similarity percentage, highlighted matches, and an AI score that only displays when the institution switches it on [1]. Because you only ever see exactly what that report contains if you preview it before you submit, understanding the mechanics first is the practical first step.
What Does Turnitin Check When It Scans a Submitted Paper?
Turnitin's core job is similarity checking: every submitted paper is compared against continuously updated indexes that include public web content, scholarly publications, and millions of student papers previously submitted through the platform [2]. The Similarity Report then shows an overall percentage that reflects how much of the document matches those sources, alongside color-coded highlights that let you see each match at a glance [2].
The percentage is not a verdict — it is a starting point for review. Quoted material and bibliographies can be excluded from the calculation, and every match is linked to the specific source it came from, so an instructor can quickly judge whether a similarity is properly cited or genuinely problematic [2]. Because the index is continuously refreshed, two checks run at different times can surface slightly different results.
When an institution enables it, Turnitin also evaluates the document with its AI writing detection model in the same submission workflow. The AI indicator is a separate score from similarity, and it appears only when the instructor has turned it on — many universities configure this per assignment rather than campus-wide [2].
Finally, the report consolidates everything into one interface that instructors open from the learning management system: the similarity score, match highlights, filters, and any enabled AI indicator. That single view is exactly what your own preview should mirror if you want to know what a professor will see [2].
How Accurate Is Turnitin's AI Writing Detection, and What Does an AI Report Look Like?
Turnitin's AI detection is a sentence-level probabilistic model trained on a large corpus of academic writing, which is why it is calibrated to student prose rather than generic internet text [3]. When a paper is scanned, the model scores each sentence and estimates the probability that it was generated by an AI system, then aggregates those scores into the overall AI percentage shown in the report [3].
On accuracy, Turnitin publishes a false-positive rate of under one percent for documents containing enough long-form prose, and it explicitly positions the indicator as evidence for instructor judgment rather than a standalone accusation [3]. Short texts, heavily edited text, and non-prose content such as bullet lists can reduce confidence, which is why the report shows an AI score only when the document meets minimum length and prose requirements.
Visually, the AI writing report mirrors the similarity report's structure: an overall AI score at the top, sentence-by-sentence highlighting that labels segments as "AI written" or "AI-paraphrased," and a plain-language note that the percentage is an estimate, not a certainty [3]. In the report display students typically see, any score below 20 percent is shown as an asterisk (*%) rather than a single-digit number, so the only explicit low numeric outcome you will encounter is 0 percent.
The key takeaway is that the number is directional, not definitive. Turnitin advises using the indicator together with the full document, the instructor's knowledge of the student's writing, and a conversation with the student before any conclusions are drawn [3].
How Can You Preview Your Turnitin AI and Similarity Report Before Submitting to Your Instructor?
Because the detection model is probabilistic and trained on academic writing, the score it assigns depends on patterns — word choice, sentence structure, and consistency — that you can inspect before anyone else does [4]. Understanding the model behind the flag is what makes a pre-submission preview useful: you are not guessing, you are looking at the same evidence the instructor will see [4].
In practice, the instructor's view is a report interface that combines the similarity percentage, match highlights, and the AI indicator in one place, with the AI indicator enabled at the assignment level. The most reliable way to preview your own version of that report is to run your draft through the same report format — similarity and AI together — before the submission deadline [4].
A preview also protects you from the two surprises that cause most post-submission problems: an unexpectedly high similarity percentage from a commonly used source, and an AI flag on text you believed was original. Both issues are fixable before submission and nearly impossible to walk back afterward [4].
Finally, choose a preview that mirrors the institutional report rather than an approximation, and confirm your draft is not archived or sent to any third-party database when you check it. A true preview gives you the same report your professor sees — without the consequence of a submission you cannot retract [4].
Knowing how Turnitin scores similarity and AI writing is only half the work — the valuable part is seeing your own paper's report before your instructor does. turnitin0 is built for exactly that moment: upload your.docx,.pdf, or.txt file and receive a real Turnitin AI writing report and similarity report — the same reports university professors see in their academic systems — delivered within minutes, with no subscription and no archive of your paper. Preview your draft now, and go into submission day with zero surprises.
FAQ
Does Turnitin check plagiarism and AI writing in the same scan?
Yes. The similarity engine and the AI writing detection model run as part of the same submission workflow, but they produce separate indicators, and the AI indicator only displays when the instructor enables it at the assignment level [1][2].
What is considered a good similarity score on Turnitin?
There is no universal pass/fail threshold; each institution or instructor sets its own expectation. The report's color-coded highlights matter as much as the percentage, because properly cited matches are a normal part of academic writing [2].
How accurate is Turnitin's AI detector?
Turnitin reports a false-positive rate of under one percent for documents with enough long-form prose, and it positions the indicator as a prompt for instructor judgment rather than proof of misconduct on its own [3].
Can I see my Turnitin AI score before I submit to my instructor?
Only if your institution grants report access or you check your draft with a service that produces the same report format. Previewing before submission is the only way to know your AI and similarity scores in advance [4].
Does checking my paper with a third-party service add it to Turnitin's database?
No. Services like turnitin0 do not archive submitted papers or share them with any third-party database, whereas a formal submission through your university is added to the repository for future similarity checks [1].