Yes, Turnitin includes an AI writing indicator that estimates how much of a submission was likely generated by AI, shown to instructors as a percentage with highlighted passages. It's a probability-based signal, not proof, and Turnitin itself has said it shouldn't be the only basis for a misconduct decision. How much weight it carries depends on your institution's policy and your instructor's judgment.
ZeroTell publishes this blog and makes its own AI detector. Turnitin is a separate company selling to schools; the details below come from its public statements and documentation.
What the AI writing indicator is
Turnitin is best known for similarity checking: comparing papers against web pages, publications and past submissions. In 2023 it added a separate feature that looks for AI-generated writing.
The indicator works differently from similarity checking. Instead of matching your text against a database, it reads the text in segments and estimates how likely each segment is to have been produced by a large language model. The output is:
- an overall percentage of qualifying text the model considers likely AI-generated, and
- highlighted segments showing where those passages are.
Turnitin has also said its model tries to identify AI text that was later run through paraphrasing tools, and it has updated the model over time. Because detectors change, any specific claim about what it can or can't catch today may be out of date tomorrow.
For the specific case of ChatGPT text and edits, see can Turnitin detect ChatGPT?
Who sees the AI score
This is one of the most common questions, and the answer surprises students. In typical setups, the AI writing report is visible to instructors and administrators, not to the student who submitted the work. Your institution's configuration decides the details.
That means you may not know your score unless your instructor shares it. If you're curious or concerned, it's reasonable to ask.
Known limits Turnitin and others acknowledge
AI detection is probabilistic. That's not a criticism unique to Turnitin; it's true of every detector. Some limits are widely discussed:
- False positives. Human writing can be flagged. Turnitin has acknowledged this and said low percentages are less reliable, so it presents them with extra caution in the report.
- Short text. Detectors need enough words to judge patterns. Very short submissions are harder to assess, and Turnitin sets a minimum length for the indicator.
- Formulaic writing. Lab reports, legal summaries, and templated assignments naturally read as predictable.
- Non-native English writers. Research and educator reports have raised concerns that simpler, more regular English can look more "AI-like" to detectors.
- Mixed and edited text. When people and AI both shape a document, results get harder to interpret.
How institutions actually use it
There's no single approach. Broadly, schools fall into a few camps:
| Approach | What it looks like | Why schools choose it |
|---|---|---|
| Indicator off | AI writing report disabled | Concern about false positives and fairness |
| Signal for conversation | Instructors may view it but must investigate further | Balance between awareness and caution |
| Formal evidence among others | Score included in an integrity file with drafts and interviews | Structured process with multiple checks |
| Course-level choice | Each instructor decides | Varying assignments and AI policies |
Several universities have publicly announced they disabled or limited the AI indicator, citing reliability concerns. Others have kept it with guidance that it must not be used alone.
What a careful instructor does with a high score
A responsible review usually includes:
- Reading the highlighted passages to see whether they actually feel off.
- Comparing with the student's other work, including in-class writing.
- Checking citations, since AI often produces sources that don't exist.
- Talking with the student about the content and process.
- Asking for drafts or version history.
What students should expect
If you're asked about a score, you'll usually get a chance to explain. The process, timelines and appeal options should be in your school's academic integrity policy. Read it before you need it.
Reading the report: a short example
Imagine an instructor opens a 1,800-word history essay and sees a moderate AI percentage, with two paragraphs highlighted near the end.
A rushed reading says "this student cheated." A careful reading asks different questions. Are the highlighted paragraphs the conclusion and a summary of background facts, which are naturally generic? Do they make specific points tied to the course readings, or broad statements that could appear in any essay on the topic? Does the citation style suddenly change there? Does the student's version history show those paragraphs typed out over time, or pasted in at once?
Sometimes the answer points to AI use. Sometimes it points to a student who wrote a bland conclusion at 1 a.m. The indicator flagged where to look; the judgment still needs a person who knows the course and the student.
If you're a teacher using the indicator
The AI score is most useful as a prompt to look closer, not as a conclusion. A few practices help:
- State your AI policy clearly in the syllabus, including what's allowed and whether disclosure is required.
- Design assignments that show process: proposals, outlines, drafts, reflections.
- Know your students' voices through low-stakes in-class writing early in the term.
- Treat low percentages with extra caution, as Turnitin itself advises.
- Have a conversation before an accusation. Many cases resolve with a discussion.
For students: protecting yourself
Whether or not your school uses the indicator, these habits help:
- Know the rule for every course. AI policies can differ between instructors.
- Write in documents with version history, such as Google Docs or Microsoft 365.
- Keep notes and sources organized as you research.
- Disclose AI help when your course asks you to.
- Don't try to game the score. Running text through tools to change how it reads to a detector is treated as concealment in many policies, and it rarely ends well.
If your writing is flagged and you didn't use AI, gather your evidence and ask for a meeting. Our article on AI detection false positives explains what to bring.
Similarity, AI writing, and what each catches
It helps to keep the two Turnitin reports separate in your head:
- Similarity report: "Does this text match something that already exists?" It catches copying and poor citation.
- AI writing indicator: "Does this text read like it came from a language model?" It looks at writing patterns, not sources.
A student can plagiarize without AI (copying from a website) or use AI without plagiarizing an existing source (new AI-generated text). That's why schools use both, and why neither alone tells the whole story.
Key takeaways
- Turnitin's AI writing indicator estimates the share of text likely written by AI and highlights those passages.
- It's usually visible to instructors, not students, depending on institutional settings.
- It's probabilistic: false positives happen, especially with short, formulaic, or non-native writing.
- Schools vary widely, from turning it off to using it as one piece of evidence.
- Your best protection is following your course's AI policy and keeping a clear record of how you wrote.
Next step: find your school's academic integrity policy and read the section on AI. It's usually shorter than you expect.