Does Turnitin detect AI? How the score works and its limits

How does Turnitin detect AI writing?

Turnitin scores each sentence and reports the share it labels as AI. Its AI writing FAQ says the model gives every sentence a score between 0 and 1, judged in context with the text around it. Sentences above a set threshold are labeled AI. The document score is the percentage of evaluated sentences classified as AI, and text the model leaves out does not count.

The same FAQ says its transformer-based model is “not explicitly programmed” to measure signals such as “burstiness” or “perplexity”. It learns statistical patterns from training data instead. Our guide to how AI detectors work covers the general idea.

AI detection is separate from the similarity (plagiarism) check. Turnitin says the AI percentage is “different from and independent of the similarity score”, and AI highlights do not appear in the Similarity Report. Our AI detector vs plagiarism checker guide explains how to read the two side by side.

Not every school gets AI scores. AI writing detection is only available to customers that license Turnitin Originality, and administrators can switch it off. In Canvas, Moodle or Blackboard, there is no AI score in the learning platform itself; instructors open Turnitin's report to see it. Our guide to which AI detector your school uses covers each platform.

What does Turnitin flag, and what does it skip?

Turnitin checks long-form prose in seven languages and leaves most other content out. The table sums up what Turnitin's own guides say.

Mixed files need care. Because only qualifying prose counts, Turnitin warns that a document with several writing types can show a percentage that does not match its highlights. In documents of only a few hundred words, the prediction is “mostly all or nothing”, so a mix of AI and original text can be flagged as entirely AI-generated.

Does Turnitin detect ChatGPT, Claude and Gemini?

Yes, according to Turnitin. Its FAQ lists the models its English detector can detect. As of November 2026 the list includes GPT-4o, GPT-5, GPT-5.2 and GPT-5.4; Claude Sonnet 4.5, Claude Sonnet 4.6 and Claude Opus 4.6; and Gemini 2.5 Pro and Gemini 3 preview models. It also lists Grok, Llama, DeepSeek, Mistral and Amazon Nova models.

Turnitin says it is “continuously updating” the model for new language models. Its model release notes say old reports are not updated; a file must be resubmitted to get a newer score. Those notes and Turnitin's release notes say Spanish, Japanese, Arabic, German, Dutch and Turkish submissions are checked by separate models.

Can Turnitin detect AI paraphrasing and humanizers?

Turnitin says yes, for English. In December 2023 it began to detect likely AI-generated text “even if it may have been paraphrased using an AI word spinner”. Turnitin names Quillbot as an example of a paraphrasing tool or word spinner; our guide to QuillBot's AI detector covers that tool separately.

In August 2025, Turnitin announced AI bypasser detection, aimed at text that “may have been intentionally modified by AI humanizer tools”. Its FAQ describes bypassers as tools, “also called humanizers”, that modify AI text to evade detection. Three limits apply:

• Paraphrase and bypasser detection is only available for English submissions.
• Beyond citing Quillbot as an example, Turnitin does not list the paraphraser and bypasser tools it was trained to detect. It says a list “would make it easier for students to evade our system”.
• The report no longer separates paraphrased text. In August 2026, Turnitin merged its purple AI-paraphrased category into a single blue “AI-generated” category. It says the model still detects text modified by paraphrasers or bypassers.

For more on how rewriting tools relate to Turnitin, see our page on Turnitin and humanized writing. Whatever the tool, rewriting AI text does not make it allowed if your course bans it. Our responsible AI use guide covers permission and disclosure.

Does Turnitin flag Grammarly as AI?

Usually not for grammar fixes. Turnitin says that in its tests on human-written documents, changes made by Grammarly “in most cases” were not flagged as AI-written, and the same went for other grammar checkers. Text made with Grammarly's generative features, such as draft generation, paraphrasing and summarizing, “will likely be flagged”. Our guide on whether Grammarly shows up as AI goes into the details.

What do the Turnitin AI score and colors mean?

The AI indicator has a few states, and only some show a number. Turnitin's guides to accessing and using the AI Writing Report describe them like this:

• Blue with a percentage: the file processed. The number is the share of qualifying text Turnitin identified as likely AI-generated, including text it thinks a paraphraser or bypasser changed.
• 0%: no qualifying text was identified as likely generated or altered by an AI tool.
• Asterisk (*%): AI was detected below 20%. No exact percentage or highlights are shown, because results in the 1% to 19% range are less reliable and false positives are more likely. Reports made before July 8, 2024 may still show a number below 20%.
• Gray with no percentage: the file could not be processed, for example because it predates AI detection or does not meet the file requirements.
• Red error icon: the submission failed to process. Turnitin suggests trying again.

Inside the report, likely AI-generated text is highlighted in blue. Purple highlights for AI-paraphrased text were retired in August 2026 for newly submitted files. The percentage also “is not necessarily the percentage of the entire submission”, because text that is not long-form prose is left out.

How accurate is Turnitin's AI detection?

Turnitin says its document false positive rate is under 1% for documents with over 20% AI writing. In its FAQ's words, it “might flag a human-written document as AI-written for one out of every 100 fully-human written documents.” It says it tests every update on over 700,000 academic papers written before ChatGPT was released.

Turnitin also says it accepts missing some AI text to keep false positives low. Its example: if it identifies 50% of a document as likely AI-written, “it could contain as much as 65% AI writing.” Errors were higher for single sentences when Turnitin last published a figure. In a June 2023 post, Turnitin put its sentence-level false positive rate at “around 4%”, most often where human and AI text meet. Its FAQ says an October 2026 model update targeted those transition points; it gives no newer sentence-level figure.

Turnitin lists what can lead to false positives: “content without a lot of structural variation, text that literally repeats itself, or text that has been paraphrased without developing new ideas.” Our list of signs of AI writing shows patterns readers notice too.

Independent research is more cautious. A 2023 study by Weber-Wulff and colleagues tested 14 detection tools, including Turnitin, and concluded that “the available detection tools are neither accurate nor reliable”. It found that the tools lean towards classifying text as human-written rather than detecting AI text, and that content obfuscation techniques significantly worsen their performance. That study tested 2023 versions, and Turnitin has updated its model several times since.

On non-native English writers, a 2023 study by Liang and colleagues found that several GPT detectors “consistently misclassify non-native English writing samples as AI-generated”. Turnitin points out that its detector was not among the tools tested. Its own 2023 study found no statistically significant difference in false positives between native and non-native English writers, for documents of 300 words or more.

At least one university has turned the feature off. Vanderbilt University disabled Turnitin's AI detector in August 2023. It noted that it submitted 75,000 papers in 2022, so at a 1% false positive rate around 750 could have been wrongly labeled as partly AI-written.

Can students see their Turnitin AI score?

No, not directly. Turnitin says the AI writing indicator and report “are not visible to students”. Instructors can download the AI report as a PDF and share it with you. You also cannot buy Turnitin to check your own work: its subscriptions “are designed for educational institutions, not for individual purchase”.

What should you do if Turnitin flags your work?

Treat a flag as the start of a conversation, not a verdict. Turnitin says its model “may misidentify human-written, AI-generated, and AI-paraphrased text” and should not be the sole basis for adverse actions against a student. Its guidance for a high percentage starts with the institution's policy and says the final decision on misconduct rests with the instructor.

1. Ask what was flagged. Ask for the AI report PDF and which passages concern your instructor. Note whether the result was a percentage or an asterisk.
2. Gather your process evidence. Collect drafts, notes, outlines, sources and version history. Do not create “drafts” after the fact.
3. Explain how you wrote the flagged parts. Describe any tools you used accurately, including grammar checkers and translation.
4. Read the policy and the appeal route. Check your syllabus and your school's academic integrity policy for what is allowed and how reviews work.

For teachers, Turnitin says the score is used as intended when it is treated as “a single data point rather than a definitive response”. Our guide to AI detector false positives explains how to show you wrote your work and includes an appeal template.

Common questions

Sources

Checked on November 10, 2026. Turnitin pages describe Turnitin's own product, and the figures in them are Turnitin's claims.

• Turnitin: AI writing detection capabilities FAQs
• Turnitin: Using the AI Writing Report
• Turnitin: How to access the AI Writing Report
• Turnitin: File requirements for an AI Writing Report
• Turnitin: File requirements for submitting your assignment
• Turnitin: AI writing detection model release notes
• Turnitin: Release notes (August 4, 2026 AI Writing Report update)
• Turnitin: What should I do if the AI writing percentage is high?
• Turnitin: How should I review the AI Writing report?
• Turnitin: How to purchase a Turnitin subscription
• Turnitin: AI bypasser detection press release (August 27, 2025)
• Turnitin: Understanding the false positive rate for sentences (June 2023)
• Turnitin: Study of bias against English Language Learners (October 2023)
• Weber-Wulff et al.: Testing of detection tools for AI-generated text, International Journal for Educational Integrity (2023)
• Liang et al.: GPT detectors are biased against non-native English writers (2023)
• Vanderbilt University: Why we're disabling Turnitin's AI detector (August 2023)