How to Read a Turnitin AI Report: Scores, Highlights & Access
Learn who can access the AI Writing Report, what 0% and *% mean, and why editing highlighted passages does not guarantee a lower result.
AI-generated conceptual illustration
You can open a Turnitin Similarity Report and still have no access to the AI Writing Report. They are separate assessments. The first compares text with sources; the second estimates likely AI writing. A green similarity score does not certify human authorship.
This guide explains the AI report’s overview, percentage, highlighted passages and practical limits.
Who Can Get the Turnitin AI Report?
In the usual institutional Feedback Studio workflow, the AI report is accessed on the instructor side when the institution has the relevant license and feature enabled. Students should ask their teacher for the result or an explanation; access to a Similarity Report does not automatically give access to the AI report. Permissions and integrations can differ, so your institution should confirm the arrangement for your course.
The official access guide explains how staff open AI Writing from the enhanced report’s tab or the classic report’s indicator. Turnitin’s Teams FAQ specifically limits AI detection in that integration to instructors.
If you need a pre-submission AI check, contact Purply. We can help arrange a Turnitin AI check and interpret the report. This is a separate check, not access to your university’s submission record, and it cannot guarantee what a later institutional check will show.
First Confirm Which Report You Have
| Heading or field | What it describes |
|---|---|
| Overall Similarity / Integrity Overview | Text matches and similarity review |
| AI Writing Overview / detected as AI | AI writing assessment |
| AI Writing Submission | Submitted text shown with AI assessment highlights where available |
For source matches, read the Similarity Report guide. For blue, green, yellow, orange, red or purple in a similarity interface, read the Turnitin colour guide.
How to Read the AI Writing Overview
Start with the document version and report date. A result belongs to that submission, not to every version of the essay. Next read the overall percentage, category labels and review notice together.
The percentage concerns eligible prose assessed by the detector. It is not the probability that a student used AI, and it need not cover every word in tables, references or other non-prose material. Category counts should be read with their labels, not treated as a count of misconduct incidents.
Current English reports use a unified AI-generated category, including likely AI-generated text subsequently altered by AI tools. Do not assume that a separate AI-paraphrased percentage will appear. Older exports can look different. Turnitin explains the consolidation in its product updates.
What Do 0%, *% and a Visible AI Percentage Mean?
| Result | How to read it |
|---|---|
| 0% | The processed report did not identify eligible text as likely AI writing; it is not proof of human authorship. |
| *% | A suppressed low result in the 1–19% range, not zero or a processing failure. |
| 20–100% | A visible estimate for assessed prose, requiring interpretation rather than an automatic decision. |
| Loading, unavailable or error | A report-state issue; no result should be assumed. |
Turnitin suppresses exact percentages and highlights for the low non-zero range. Its AI report documentation also warns that detection can make mistakes and needs human review. The review notice is not itself a finding against the student.
If the result is missing, use the separate AI score troubleshooting guide.
How to Read Highlighted Text
Look at the heading and legend first. In the current English AI report, blue/cyan shading identifies text assessed as likely AI-generated, including text likely revised by AI tools. It is not the blue similarity-score band and is not a matching-source number.
Read each highlighted paragraph alongside the surrounding argument, notes and drafts. A highlight is a model assessment, not evidence of the exact tool used or a verified history of how the sentence was written. An unhighlighted sentence is not certified human-written.
If a PDF has no AI highlights, check its result and report state. A processed 0% result, a suppressed low result and an unavailable report are different situations; blank highlighting alone does not explain which one applies.
Why Can Different Paragraphs Be Flagged After Editing?
In our work helping clients interpret AI checks, we have repeatedly seen this pattern: a writer edits the highlighted passages, runs another check, and finds that other passages which were previously unhighlighted are now flagged. The overall result can remain high even though the original highlighted passages have changed.
This is a service observation, not a controlled experiment or a published accuracy estimate. It does not establish whether whole-document context caused the changed highlights, or reveal the detector’s internal decision process. The practical lesson is narrower: the previous highlight map is not a fixed list of sentences whose replacement will guarantee a lower result.
A new check evaluates the submitted version again. Compare the full revised document and report, not just the original marked sentences. Do not infer a guaranteed result from isolated sentence edits or from a different detector. We cannot determine the internal cause of a changed highlight map from the PDF alone.
What to Do With a Result You Do Not Understand
- Confirm that the report belongs to the intended file and version.
- Keep the complete report and distinguish AI results from similarity matches.
- Review the writing itself: argument, evidence, citations and the way each paragraph develops your reasoning.
- Preserve outlines, research notes, source annotations and draft history. Follow your course’s AI disclosure rules where applicable.
- If the result is disputed, ask the instructor to explain the passages of concern and the review process. Provide your writing-process evidence.
Changing wording solely to chase a detector score can weaken an essay without settling the underlying issue. Use feedback to improve the work and document how it was produced. Neither a 0% report nor a pre-check replaces the assignment’s rules.
Need a Turnitin AI Check Before Submission?
Purply offers AI checking and help reading the resulting report. Tell us whether you need an AI Writing Report, a Similarity Report, or both, so the service matches your question. We can explain the overview and flagged passages and discuss writing feedback. We do not promise a particular score or a university outcome.
Frequently Asked Questions
Can students get the Turnitin AI report directly?
In the usual institutional workflow, ask your instructor for the staff-side result. Student access to similarity results does not automatically include AI results; confirm your course’s permissions.
Is a green Turnitin score an AI pass?
No. Green commonly refers to a similarity indicator. AI writing detection is separate.
Does 0% prove my paper is human-written?
No. It means the processed check did not flag eligible text as likely AI writing in that version.
Will editing only the highlights remove the AI result?
There is no guarantee. In our service experience, subsequent checks can flag previously unmarked passages. This observation does not establish the detector’s internal mechanism.
Why does the percentage not match the amount of shaded text I see?
The AI percentage concerns eligible prose, not necessarily the complete file. Check the report version, labels and assessed text before comparing it with total word count or page area.
Can Purply help me get and understand a report?
Yes. Contact us for an AI check and report interpretation. A pre-check does not guarantee a later institutional result.
Need an AI Writing Report?
Contact Purply for an AI check and help interpreting the result. Tell us whether you need AI detection, similarity checking, or both. A pre-check cannot guarantee a later institutional result.
Here's how we'd coach you through it:
Send us your draft or your question. We'll walk you through:
- Turnitin AI checking
- AI and similarity report interpretation
- Academic writing feedback