Is a Turnitin AI score proof that I cheated?

No. Turnitin says its AI score should not be used as the sole basis for adverse actions against a student, and that it does not make determinations of misconduct. Your instructor and institution decide, using their policies and other evidence.

What does the asterisk on a Turnitin AI score mean?

It means AI was detected in under 20% of the qualifying text. Turnitin says false positives are more likely in that range, so since July 8, 2024 it shows an asterisk with no score or highlights instead of a number.

How many words does Turnitin need to check for AI writing?

Turnitin's file requirements page lists at least 300 words of prose in a long form format and no more than 30,000 words, in English, Spanish or Japanese. Files that don't qualify show a gray indicator with no score.

Does Turnitin flag text that an AI tool paraphrased?

Turnitin's guide says its English detector has a separate category for text that was likely AI generated and then changed by an AI paraphrasing tool or word spinner, highlighted in purple. Its Spanish and Japanese detectors don't include that category yet.

Does Turnitin Detect ChatGPT? What Its AI Report Shows

Q: Does Turnitin detect ChatGPT?

Turnitin's AI writing detection is built to flag text likely produced by generative AI tools, including chatbots such as ChatGPT. It reports an estimated share of qualifying prose, not a yes or no verdict, and Turnitin says the result may not always be accurate.

What does Turnitin's AI report measure?

Turnitin's help guide on AI writing detection in the classic report view says the feature is designed to help educators identify text that might be prepared by a generative AI tool, including large language models, chatbots and word spinners. It does not name one chatbot. It looks for patterns of AI generated text in general, which covers drafts from ChatGPT.

The percentage it shows is easy to misread. According to the same guide, it is the share of qualifying text that the model determines was generated by AI, not necessarily the share of the whole submission. Qualifying text means prose sentences in long form writing, such as an essay or article. Turnitin says the model does not reliably detect AI text in poetry, scripts or code, and does not detect short form writing such as bullet points, tables or annotated bibliographies.

The report then splits flagged text into two categories, per Turnitin's guide:

AI generated only: text likely generated by a large language model, highlighted in cyan.
AI generated text that was AI paraphrased: text likely generated by AI and then changed by an AI paraphrasing tool or word spinner, highlighted in purple. Turnitin says only its English detector includes this paraphrase category for now.

Turnitin also says the AI percentage is separate from, and independent of, the similarity score. They answer different questions: one is about matching sources, the other about likely AI generation.

When does Turnitin produce an AI score at all?

Not every file gets one. Turnitin's page on file requirements for an AI writing report lists the conditions:

Requirement	Turnitin's published limit
Minimum length	At least 300 words of prose in a long form writing format
Maximum length	No more than 30,000 words
Languages	English, Spanish, Japanese
File types	.docx, .pdf, .txt, .rtf
File size	Less than 100 MB

These are the limits shown on Turnitin's guide when we read it on 8 November 2026. Turnitin has changed them before, so check the page if your submission is close to a limit.

What do instructors see in the report?

The AI writing indicator sits in the side panel of the Similarity Report. The classic report guide describes the states it can show:

What the indicator shows	What Turnitin says it means
Blue, with 0% or a score from 20% to 100%	Processed. The percentage is the share of qualifying text the model judged AI generated
An asterisk (*%)	AI was detected below 20%. No score or highlights are shown because false positives are more likely in this range
Gray, with two dashes	Not processed, for example because the file did not meet the requirements
An error symbol (!)	Turnitin failed to process the submission and the report cannot be opened

The asterisk is worth understanding. Turnitin's guide says its testing found a higher incidence of false positives when the percentage is between 1 and 20, so for reports generated from July 8, 2024 onward it shows an asterisk instead of a number in that range. Older reports may still show a number below 20%.

Instructors can open the full AI writing report to see the highlighted sentences and an interactive breakdown bar. What they do next is not decided by Turnitin. That depends on the course, the instructor and the institution's academic integrity policy.

How often does Turnitin report false positives?

Turnitin's most detailed public figures on false positives come from a May 2023 post by its Chief Product Officer, AI writing detection update from Turnitin's Chief Product Officer. In it, Turnitin said:

For documents with over 20% AI writing, its document false positive rate was less than 1%, validated by a test of 800,000 academic writing samples written before ChatGPT's release.
Its sentence level false positive rate was about 4%, meaning a highlighted sentence may be human written about 4 times in every 100 highlighted sentences.
False positives were more common in the first and last few sentences of a document, and sentence level false positives were more common in documents that mix human and AI writing.
It raised the minimum length for an AI score from 150 to 300 words, because accuracy improved with more text.
These metrics may change as models evolve.

Those are Turnitin's own numbers from 2023, measured on its own tests, so read them as vendor figures. Turnitin's earlier post, Understanding false positives within our AI writing detection capabilities from March 2023, defines a false positive as fully human written text incorrectly identified as AI generated. It also adds the part that matters most for students: Turnitin does not make a determination of misconduct, and because its false positive rate is not zero, instructors need to apply professional judgment and their institution's policies.

Independent researchers have raised a related concern about AI detectors in general. A 2023 Stanford study, Liang et al., "GPT detectors are biased against non-native English writers", found that the detectors it tested often misclassified essays by non-native English writers as AI generated. That study did not test Turnitin, but it is worth knowing if English is your second language. Our guide on how to tell if something was written by ChatGPT explains why text based signs are weak evidence on their own.

What should you do if Turnitin flags your work?

A flag is the start of a conversation, not a verdict. Turnitin's own guide says the score should not be used as the sole basis for adverse actions against a student. If your work is flagged:

Stay calm and ask what was flagged. Ask your instructor which passages were highlighted and what the score was, including whether it was an asterisk result.
Gather your process evidence. Drafts, outlines, notes, research, and the version history in your document editor show how the work developed.
Bring your ChatGPT history if you used it. OpenAI's educator FAQ suggests teachers ask students to share their ChatGPT conversations, which show how the tool was used. If your use was allowed, the shared conversation is your best evidence.
Explain your argument in person. Being able to talk through your reasoning and sources is strong evidence the thinking is yours.
Follow the formal process if it goes further. Read your institution's academic integrity policy, including how to respond or appeal, and ask a student advisor or ombudsperson for help if you have one.

Turnitin publishes a guide for exactly this situation, AI conversations: Handling false positives for students, dated 6 May 2026, with strategies to use before writing and after submission plus a self reflection form.

How do you use ChatGPT within your course rules?

The safest approach is to treat ChatGPT like any other source of help and follow the rules you were given.

Read the policy first. Check the syllabus, the assignment brief and your institution's AI guidance. Rules differ between courses, and even between assignments in the same course.
Ask when unsure. A one line email asking "Can I use ChatGPT to brainstorm or outline for this essay?" settles it.
Disclose when asked. Say which tool you used and for what, in the format your instructor wants.
Keep your own argument and sources. Use ChatGPT for permitted help such as brainstorming or feedback, and write the claims and citations yourself.
Keep records. Save your drafts and share links to your ChatGPT conversations, so you can show your process if asked.

If your course allows AI assisted drafting and you are editing a paragraph so it reads in your own voice, our ChatGPT essay humanizer opens with an academic tone and a disclosure checklist, and the main ChatGPT Humanizer box handles any short draft. For longer academic documents you wrote yourself, such as a thesis chapter or a report beyond the box's 1,000 word limit, one option built for long documents is Walter Writes. Whatever you use, your course rules on AI use and disclosure still apply.

Curious whether ChatGPT can do this editing job itself? Our guide, can ChatGPT humanize its own text, looks at where self rewriting helps and where it stops helping.