What Is Turnitin? Similarity and AI Reports Explained

Turnitin is an academic-integrity platform that compares submitted writing with internet pages, publications, and previously submitted papers. Depending on an institution's settings and products, it can produce a Similarity Report, an AI Writing Report, feedback, and grading tools. Turnitin supplies evidence for review; it does not itself decide whether plagiarism or misconduct occurred.

Turnitin at a Glance

What Does Turnitin Check?

Turnitin's Similarity Report compares text in a submission with the databases selected for an assignment. Turnitin's current documentation describes three broad source groups:

• active and archived internet pages;
• previously submitted student papers;
• periodicals, journals, and other publications.

The report highlights matching or highly similar passages and links them to possible sources. Correct quotations, references, common phrases, templates, and a writer's own earlier submission can all create matches. That is why Turnitin's similarity-score guide says it checks for similarity rather than making a plagiarism determination.

Some Turnitin configurations also produce an AI Writing Report. That report evaluates qualifying long-form prose and identifies text the model considers likely AI-generated or likely AI-generated and then AI-paraphrased. It is separate from the Similarity Report, uses different eligibility requirements, and may not be enabled for every user or assignment.

How Does Turnitin Work?

The exact workflow varies because institutions can use Turnitin through its website, a learning-management system, Google Classroom, or other integrations. The common sequence is:

1. An educator configures an assignment. Settings determine available reports, search repositories, exclusions, student visibility, due dates, and whether resubmission is allowed.
2. A student submits a document. Turnitin checks whether the file meets the relevant assignment and report requirements.
3. Turnitin processes the text. The similarity system compares the submission with selected sources. If AI detection is enabled and the prose qualifies, the AI writing model performs a separate classification.
4. Reports become available. Processing can take time. Student access to a Similarity Report depends on the educator's settings; AI-report access may be restricted.
5. A person reviews the evidence. The reviewer examines individual matches, quotations, references, assignment instructions, drafts, and institutional policy before drawing a conclusion.

Turnitin's student submission guide recommends confirming the correct file and saving the digital receipt and submission ID. Resubmission rules also depend on assignment settings, so students should not assume that uploading a new file will always be available.

What Is the Turnitin Similarity Score?

The Similarity Score is the percentage of a submission's text that matches sources in Turnitin's selected databases. It is not a probability of plagiarism, a grade, or a measure of how much of the paper is “original.”

For example, a well-cited literature review may contain many legitimate matches, while an uncited copied sentence could be important even when the overall percentage is low. The useful questions are:

• Which passages matched?
• What sources produced those matches?
• Are quotations and paraphrases attributed correctly?
• Did a template, bibliography, or prior draft inflate the result?
• What does the assignment require?

Turnitin provides colored ranges in parts of its interface, but colors do not establish universal pass/fail thresholds. Its help center states that there is no acceptable percentage that applies to every paper. The instructor or institution decides how the report fits its policy.

What Is the Turnitin AI Writing Percentage?

The AI writing percentage represents the share of qualifying text that Turnitin's model identifies as likely AI-generated or likely AI-generated and then modified using an AI paraphrasing or bypasser tool. It is not the chance that a person used ChatGPT, and it is independent of the Similarity Score.

Turnitin's current AI Writing Report guide says qualifying text means prose sentences in long-form writing such as an essay, dissertation, or article. The AI report currently has stricter file, language, and word-count requirements than the Similarity Report.

Turnitin also says the model can misidentify human-written, AI-generated, and AI-paraphrased text. Scores above 0% and below 20% are shown as an asterisk without an exact percentage or highlights because false positives are more likely in that range. Even a surfaced result requires further scrutiny and human judgment.

For a detailed explanation of the classifier and report categories, read the Turnitin AI detection guide.

Similarity Report vs AI Writing Report

Confusing these reports is a common error. A paper can have a high Similarity Score because it quotes sources correctly and no surfaced AI result. Another paper can have little source overlap but receive a separate AI writing indicator.

What Turnitin Does Not Tell You

Neither report establishes intent. A text match does not explain whether the writer copied, quoted, collaborated, reused authorized material, or encountered a common phrase. An AI classification does not identify a particular prompt, account, model, or writing process.

Turnitin also does not set one worldwide academic-integrity policy. Each institution decides which assistance is permitted, what students must disclose, how evidence is reviewed, and what appeal process applies. Product settings and report visibility can therefore differ between classes at the same school.

What Students Should Do Before Submitting

1. Read the assignment's AI and source-use policy. Do not infer permission from what a tool can detect.
2. Check every quotation and paraphrase. Make sure attribution and references follow the required style.
3. Keep evidence of the writing process. Preserve outlines, research notes, drafts, comments, and version history.
4. Confirm the file. Preview the uploaded document and save the digital receipt or submission confirmation.
5. Use an available draft report carefully. Review the actual matching passages instead of chasing a target percentage.
6. Ask before resubmitting. Availability and timing depend on assignment settings.
7. Discuss unexpected results. If authentic writing is questioned, provide process evidence and request review under the institution's policy.

Do not introduce errors, remove valid citations, or repeatedly rewrite authentic prose solely to lower a detector score. Those actions can reduce quality and erase evidence of authorship.

What Educators Should Review

Turnitin's reports are most useful when they start a focused review instead of ending one. Educators can:

• inspect each material match rather than relying on the overall color;
• distinguish properly cited language from unattributed reuse;
• check whether exclusions or an earlier draft affected the score;
• compare questioned writing with prior work and in-class performance;
• ask the student to explain sources, reasoning, and revision decisions;
• examine drafts and version history;
• apply a documented policy consistently;
• record the human conclusion and evidence used.

Turnitin's AI-report review guidance explicitly describes the AI score as one data point rather than a definitive answer.

Frequently Asked Questions

Is Turnitin a plagiarism checker?

Turnitin is commonly called a plagiarism checker, but its official guidance says the Similarity Report finds matching text. An educator determines whether a match reflects plagiarism, correct citation, authorized reuse, or another explanation.

What does Turnitin check?

The Similarity Report compares submissions with selected internet, publication, and prior-submission repositories. If enabled, a separate AI Writing Report classifies qualifying long-form prose for likely AI generation or AI paraphrasing.

Can Turnitin detect ChatGPT?

Turnitin says its AI Writing Report can identify qualifying prose likely generated by large language models. It cannot prove that a specific person used ChatGPT, and the model can make mistakes.

What is a good Turnitin score?

There is no universal good score. Review the matched passages, sources, citations, assignment type, and institutional policy instead of treating one percentage as a pass/fail threshold.

Can students see their Turnitin score?

Students may see a Similarity Score when the educator enables access. Report visibility and timing vary by assignment, integration, and institution.

Are the Similarity Score and AI score the same?

No. The Similarity Score measures source overlap; the AI writing percentage classifies qualifying prose. They are calculated independently and answer different questions.

Sources

• Turnitin: Understanding the similarity score
• Turnitin: Overview of the Similarity Report
• Turnitin: Using the AI Writing Report
• Turnitin: How should I review the AI Writing report?
• Turnitin: Submitting to a Standard Assignment
• Turnitin: File requirements for student submissions