Do Journal Reviewers Check Turnitin's AI Detector?

Some journals may use iThenticate or another service during manuscript screening, but you should not assume that an external peer reviewer checks Turnitin's AI detector or sees its AI Writing Report. An editorial office or editor may handle screening, while each journal decides how reports move through its workflow. Turnitin also treats text similarity and AI-writing detection as separate results, and iThenticate 2.0 customers need a separate AI-writing add-on. The only reliable answer for one submission is the target journal's written policy or a direct answer from its editorial office.

A Turnitin score you generated elsewhere cannot settle that question. It tells you what one enabled account reported for one file at one time. It does not show which product the journal licenses, whether AI detection was enabled when your manuscript entered the system, who received any report, or how the journal uses it.

This table is the main answer. The phrase "the journal checked my paper on Turnitin" is too broad to identify the report, the person who saw it, or the stage at which it was used.

Crossref's Similarity Check documentation describes editors submitting manuscripts to iThenticate, receiving a Similarity Report, and reviewing the matches before publication. That is evidence for an editor-facing similarity workflow. It is not evidence that every external peer reviewer runs the check personally or that every journal performs the check at the same point.

The distinction is about responsibility, not a guaranteed sequence. Public Turnitin and Crossref documentation does not say when every journal performs each check or which materials it later shares with reviewers. A comment from a reviewer may come from material supplied by the journal, or it may be the reviewer's own observation rather than a detector result. Ask the journal which path applied instead of inferring it from the reviewer's wording.

Current publisher policy reinforces the role boundary. Elsevier's journal policy has separate sections for authors, reviewers, and editors. It gives reviewers responsibility for their own scholarly assessment and editors responsibility for the editorial process and final decision. That policy does not establish a universal Turnitin workflow, but it is a clear reason not to treat "editor" and "reviewer" as interchangeable words.

Crossref describes Similarity Check as a way to find text overlap. Its documentation says iThenticate produces a Similarity Score and highlighted matches, which editors then interpret.

Turnitin's AI Writing Report guide draws a different boundary. It says the AI-writing percentage is independent of the Similarity Score and that AI-writing highlights are not visible in the Similarity Report itself. Turnitin's current FAQ also says iThenticate 2.0 customers can access AI-writing detection only when they license it as an add-on, and institutions can disable the feature.

So a journal can use iThenticate for similarity without generating an AI Writing Report. It may also have AI detection available without sending that output to an external reviewer. A missing AI section on a report does not identify which of those states applies.

Start with the journal's instructions for authors and the publisher's policy page. Search for "artificial intelligence," "generative AI," "screening," "similarity," and "disclosure." Then check the submission portal for any declaration you will be asked to make.

If those pages do not answer the workflow question, send the editorial office a narrow request:

Does this journal use an AI Writing Report during initial manuscript screening? If it does, is the report reviewed by the editorial office, the handling editor, external peer reviewers, or some combination of them? Which written policy governs how the result is considered?

That wording keeps three questions separate: whether the report exists, who receives it, and which rule applies. A vague question such as "Do you check AI?" can produce an answer that mixes similarity screening, disclosure review, and AI detection.

Do not treat silence on a public page as proof that no internal check exists. Public guidance can be incomplete or updated later. Record the page URL and the date you read it, and keep any reply from the journal with your submission records.

• Read the target journal's instructions for authors, not only the publisher's general policy.
• Check the submission portal early enough to see its AI-use and originality declarations before the deadline.
• Confirm whether your use of language, translation, editing, or generative tools requires disclosure under that journal's wording.
• Verify every citation, quotation, table, figure, equation, and data value in the final file.
• Save the exact submitted manuscript separately from later revisions.
• Keep dated drafts, version history, analysis files, source notes, coauthor comments, and relevant correspondence.
• If you have a report, identify its exact name. Do not call a Similarity Score an AI score.
• If a score concerns you, ask who generated the report and which policy will be applied before rewriting the whole manuscript around it.

The point of this checklist is not to manufacture a particular detector result. It is to make the submission, disclosure, and writing process clear enough to examine if a question arises.

First ask what triggered the concern. Was it an initial screening report, a comment from the handling editor, an observation by a peer reviewer, or a mismatch in the submission declaration? Those are different events.

Request the complete report if one exists, the passages under discussion, and the journal policy being applied. Preserve the submitted file before making changes. Then respond with records that fit research work: drafts, version history, data and analysis files, reference notes, tracked changes, coauthor comments, and correspondence.

Turnitin tells instructors that its model can misidentify human-written, AI-generated, and AI-paraphrased text and should not be the sole basis for adverse action against a student. That guidance is written for a classroom, not as a rule for journal publishing. The relevant takeaway is narrower: a classifier output needs human review and an applicable policy. For a journal manuscript, the journal has to tell you what that policy is.

The source boundary matters more than the vendor names. These pages were re-read on 31 August 2026:

• Crossref Similarity Check for the editor-facing iThenticate similarity workflow.
• Turnitin's AI Writing Report guide for the separation between AI-writing and similarity results, report limitations, and the classroom-facing adverse-action guidance.
• Turnitin's AI-writing detection FAQ for iThenticate add-on licensing and enablement.
• Elsevier's generative AI policies for journals for the distinction between author, reviewer, and editor responsibilities.

Each source establishes only the boundary named above. None of them says that every journal uses Turnitin's AI detector or that every external peer reviewer receives its report.

Do all journals use Turnitin or iThenticate? No universal claim is supported. Crossref documents one iThenticate-based service used by participating publishers, while each journal or publisher sets its own workflow.

If a journal checks similarity, does it also check AI writing? Not necessarily. Turnitin describes the AI capability for iThenticate 2.0 as an add-on, and the two results are independent.

Can a peer reviewer see my Turnitin AI score? Possibly, if the journal has the report and chooses to share it, but the public Turnitin and Crossref pages do not establish that every external reviewer receives it. Ask the editorial office about the access path for your submission.

Does a high score from my own Turnitin check predict rejection? No. It does not reveal the journal's licensed features, workflow, report access, or decision policy. Preserve the file and ask about the actual process rather than treating a self-run score as a decision.

Is an AI Writing Report the same as a Similarity Report? No. Turnitin says the percentages are independent, and its guide says AI highlights are not shown in the Similarity Report itself.