How to Read Your Turnitin AI Report: Indicator, Tables, Excel and What Students See

Turnitin's AI indicator scores only "qualifying text": prose sentences in paragraphs, in a file of 300 to 30,000 words. Tables, bullet points, annotated bibliographies, code, poetry and scripts are not reliably assessed, spreadsheets are not an accepted file type at all, and the report is written for instructors — students usually never see the percentage for their own work.

That one paragraph answers most of the questions people search for. The rest of this guide walks through the report the way an instructor sees it, using Turnitin's own documentation, so you know what a number means, what it ignores, and what to do about it. If your real question is whether Turnitin checks for AI in the first place, start with does Turnitin detect AI? — this page assumes the answer is yes and your university has it switched on.

The short answer

Turnitin's guide to using the AI Writing Report defines the headline number as the share of qualifying text — "prose sentences contained in a long-form writing format" — that its model judges likely to be AI-generated, or AI-generated and then run through a paraphrasing or "bypasser" tool. The same guide states that the model "does not reliably detect AI-generated text in the form of non-prose, such as poetry, scripts, or code, nor does it detect short-form/unconventional writing such as bullet points, tables, or annotated bibliographies."

So a Turnitin AI percentage is not a percentage of your file. It is a percentage of the paragraphs of continuous prose inside it. Everything else in this guide follows from that definition.

What the AI indicator percentage counts

According to Turnitin's file requirements for an AI Writing Report (checked 7 September 2026), a submission only generates an AI report if it:

• is under 100 MB;
• contains at least 300 words of prose text in a long-form writing format;
• does not exceed 30,000 words;
• is written in a supported language — the file-requirements page lists English, Spanish and Japanese, while the report guide also lists Arabic; and
• is one of the accepted file types: .docx, .pdf, .txt or .rtf.

Below 300 words of prose, no AI report is generated at all, which is why very short reflective pieces and discussion posts often show no indicator. Our guide to text length and AI detection accuracy explains why the floor exists and why short samples swing so much when other tools do score them.

The report itself, per Turnitin, has a handful of states: 0% (no qualifying text flagged), an asterisk (some text flagged but under 20% — see below), a percentage from 20 to 100, plus loading, processing-error, "file didn't meet the requirements" and "AI writing detection was not enabled at time of submission". That last state matters in the UK: the indicator is licensed and switched on per institution, and many UK universities have chosen not to run it.

Tables, bullet points and Excel: why they don't score

This is the question behind searches like "can Turnitin detect AI in Excel" and "does Turnitin detect AI in tables", and Turnitin's documentation answers it directly.

Tables and bullet points. They are outside the model's definition of qualifying text. Turnitin's report guide lists "bullet points, tables, or annotated bibliographies" among the short-form, unconventional writing the model does not detect. A table filled with AI-generated cell text is therefore not scored, and a bulleted list of AI-written points is not scored either. The words still count toward the file, but they do not enter the percentage.

Spreadsheets. An .xlsx or .csv file is not an accepted file type for the AI report — only .docx, .pdf, .txt and .rtf are. A spreadsheet submitted to a Turnitin assignment may still generate a similarity report depending on the assignment's settings, but it cannot produce an AI writing percentage. If you pasted spreadsheet output into a Word document as a table, it falls under the table rule above.

Mixed documents. Turnitin notes that "a document containing several different writing types would result in a disparity between the percentage and the highlights." In plain terms: a dissertation with 6,000 words of prose and 40 tables is scored on the prose only, so the percentage can look higher or lower than a reader eyeballing the whole document would expect. When a marker reads a report, the highlighted passages — not the headline number — show what was actually assessed.

None of this is a loophole worth planning around. Moving AI-written prose into a table to avoid the indicator produces work that reads as evasive to a human marker, and the misconduct process is a human one.

Can students see AI detection on Turnitin?

Usually not. Turnitin's report guide is addressed to instructors ("This guide helps instructors understand what the AI Writing Report shows"), and the guide states that "AI writing highlights are not visible in the Similarity Report" — the document students typically can open. Whether a student ever sees an AI percentage depends on the institution's settings and its policy on disclosure.

Three practical consequences:

1. You cannot check your own Turnitin AI score before submitting, and no third-party tool can tell you the number Turnitin would produce. An independent check such as Genutext's free AI detector can show whether your prose carries AI-like patterns and which sentences drive that — useful for understanding your own writing, not for predicting Turnitin's figure. The two systems will disagree; that is normal.
2. If you are questioned, ask to see the report. Turnitin itself says the model "should not be used as the sole basis for adverse actions against a student". A university that relies on the number without showing you the highlighted passages is departing from the vendor's own guidance, and that is a legitimate point to raise.
3. Check your university's stance first. The UK university AI-detection policy tracker records, from each university's own published statements, which institutions run the indicator and which have switched it off — with a linked source and date for every entry.

The asterisk: what 1–19% means

If the model flags some text but the total is below 20%, the report shows an asterisk (*%) rather than a number, with no highlights. Turnitin's stated reason: "there is a higher incidence of false positives when the percentage is between 0 and 19", so the score is suppressed "to reduce the likelihood of misinterpretation". This has applied to reports generated since 8 July 2024; older reports may still show a numeric score under 20%.

An asterisk is therefore not a "low AI score" to be argued over. It is Turnitin declining to assert anything. Our guide to what an AI-detection percentage means covers how to read thresholds across tools, and what AI score is acceptable at UK universities covers how institutions actually treat them.

AI-generated vs AI-paraphrased text

Since July 2024 the report splits its percentage into two interactive categories: text likely AI-generated, and text likely AI-generated and then AI-paraphrased with a word-spinner such as Quillbot. From August 2025 the AI-generated category also covers text that may have been altered by an "AI bypasser" tool. Turnitin states that only its English model includes paraphrase and bypasser detection; its Spanish and Japanese models do not.

Two things follow. First, paraphrasing ChatGPT output is explicitly within the model's remit rather than a workaround. Second, the category labels are model judgements, not evidence of which tool was used: the report cannot name the tool, the date, or whether the writer had permission to use AI at all.

Why the AI percentage differs from the similarity score

Turnitin's guide is unambiguous: the AI percentage "is different from and independent of the similarity score." Similarity measures matched text against Turnitin's databases and points to named sources; the AI indicator is a statistical classification of prose style with no source to point to. A submission can be 3% similar and 80% AI, or the reverse. The full comparison is in our does Turnitin detect AI guide, and Turnitin's own understanding the similarity score covers the plagiarism side.

What to do if your report shows a number

If you have been shown a Turnitin AI percentage — or told one exists — the number itself is the least useful part of the conversation.

• Ask which passages were highlighted. The percentage is calculated on qualifying prose; the highlights show exactly which sentences the model flagged. Those are what you need to account for.
• Bring your process, not a counter-score. Drafts, version history, notes and reading lists are evidence of authorship; another detector's percentage is not. Our guide to proving you didn't use AI sets out what actually persuades a panel.
• Know the fairness evidence. If English is not your first language, false positives are better documented for your writing than for native speakers, and it is legitimate to say so.
• Follow the procedure, not the panic. The step-by-step UK guide to being accused of using AI covers the meeting, the evidence and the appeal.

Frequently asked questions

Can Turnitin detect AI in Excel?

No. Spreadsheet files are not among the accepted file types for an AI Writing Report (.docx, .pdf, .txt and .rtf), so an .xlsx submission cannot produce an AI percentage. Spreadsheet content pasted into a document as a table is treated as a table — see below.

Does Turnitin detect AI in tables or bullet points?

Not reliably, by Turnitin's own account. Tables, bullet points and annotated bibliographies are listed as short-form or unconventional writing the model does not detect; the AI percentage covers prose sentences in paragraphs only.

Can students see AI detection on Turnitin?

Usually not. The AI Writing Report is an instructor-facing document, and Turnitin states that AI writing highlights are not visible in the Similarity Report students normally see. Whether a student is shown the percentage is decided by the institution.

What does the asterisk on a Turnitin AI report mean?

Some text was flagged but the total was between 1% and 19%. Turnitin suppresses the number and the highlights in that band because false positives are more common there. It is not a score.

Does 0% mean no AI was used?

It means the model did not flag any qualifying prose. It cannot prove absence of AI use, just as a high number cannot prove presence — Turnitin says the report should never be the sole basis for action.

Which languages does Turnitin's AI detection cover?

Turnitin's guides list English, Spanish and Japanese (with Arabic also listed on the report guide). Paraphrase and bypasser detection currently apply to the English model only.

Sources and further reading

• Turnitin: Using the AI Writing Report
• Turnitin: File requirements for an AI Writing Report
• Turnitin: AI writing detection model — release notes
• Turnitin: Understanding the similarity score
• Genutext: UK University AI-Detection Policy Tracker
• Genutext: Does Turnitin detect AI?

Turnitin is a trademark of its owner. Genutext is independent and not affiliated with, endorsed by, or connected to Turnitin. Product behaviour described here reflects Turnitin's public documentation as of 7 September 2026; check the linked pages for current details.