Short answer first: Turnitin matches your text against a database and highlights what it finds. It does not decide anything. Turnitin's own student guide says it does not check for plagiarism, and the University of Melbourne tells students the similarity score does not indicate whether plagiarism has taken place. The AI writing feature is separate, students do not see it, and it is switched off at several Australian universities: Curtin disabled it from 1 January 2026, and UQ made it unavailable for Semester 2 2025. A score is a prompt to look closer, not a finding.
What follows is what Australian universities and Turnitin publish themselves, checked in September 2026. Your unit may differ on the details, so read the assessment page as well.
What is a similarity report, and what is it not?
It is a comparison, run automatically, and the number at the top is arithmetic rather than judgement. Turnitin's guide for students is unusually direct about this: "Turnitin does not check for plagiarism". What it does is compare your submission against current and archived internet content, periodicals, journals and publications, and a repository of work other students have submitted, and highlight where your writing is similar to any of it.
The score itself is the percentage of matched text in your submission, calculated by dividing the total words by the amount matched to outside sources. Quotes and references you formatted correctly are still matches. So are the standard phrases of your discipline.
The University of Melbourne's LMS guide for students, which the page dates to 01 Jan 2025 and which is still the only Turnitin guide in its student index, says the same thing from the university's side, and says it in capitals: "the Similarity Score does NOT indicate whether plagiarism has taken place". Its page adds that a high score may contain legitimately cited text and so is not a problem, and that a low score may contain text that does cause concern. It then puts the responsibility where it belongs: teaching staff judge the quality of the report and decide whether what Turnitin flagged as non-original may reasonably be considered plagiarism.
That is the whole mechanism. A machine finds overlap, a human decides what the overlap means. The rest of this article is about what happens in between.
Which universities publish that they use it?
Enough of them that you should assume yours does unless it says otherwise, and the assumption is safer than the alternative. Melbourne, ANU, Curtin, UQ and Sydney all describe Turnitin on public pages, and the descriptions are consistent.
- ANU. Its Academic Skills page calls Turnitin an online text matching tool accessed through Canvas, and says submitted work is checked against a large database of academic publications and other sources to find matching sections.
- Curtin. Its UniSkills guide explains the report to students and describes the colours in it, telling you that while a high percentage may look concerning, it does not necessarily indicate plagiarism.
- Melbourne. Its guide notes that the software is used by many universities throughout Australia and internationally, and explains how to open, read and download your own report through Feedback Studio.
- UQ. Its eLearning project page tracks the move to Turnitin's newer similarity report, whose match categories, it says, make it easier for marking staff to identify areas to investigate and the types of referencing errors students make.
- Sydney. Its student page on artificial intelligence says that where unpermitted use is suspected the Turnitin AI detection tool may be used to evaluate the situation, and that a detector score alone would not decide the question but would be considered alongside other relevant evidence.
None of them present it as a gotcha. Every one of these pages is written as a teaching resource, which is a fair reading of what the tool is for. The assessment system it sits inside is described in how Australian university exams work.
Can you check a draft yourself?
Sometimes, and it depends on the unit rather than the university. This is the single most useful thing to find out in week one.
ANU is the clearest case. Its page says that as a student, you can use the Turnitin Practice site to check your own draft prior to submission. Curtin's guide says many units may allow you to submit a draft to Turnitin before the final submission. At Melbourne, whether you see a report at all is set by teaching staff: the page says student access to the similarity report is a preference decided by staff, including whether it appears immediately after you submit or only after the due date.
One mechanical trap catches people who leave it late. Turnitin's student guide says overwritten or resubmitted papers may not generate a new similarity report for a full 24 hours, and that the delay is automatic so a resubmission does not match against your previous draft. If you plan to submit, read the report and resubmit, you need a day, not an hour.
What to do with a report you can see
- Ignore the headline percentage for a moment and open the matched sources list.
- Check every match against your reference list, one by one.
- Look for long unbroken highlights. Those are the ones a marker will read.
- Fix quoting and citation, not the words, where the match is a real quote.
- Rewrite in your own words only where the match is paraphrase that stayed too close.
A score that drops because you cited properly is a fixed problem. A score that drops because you swapped synonyms is the same problem with a costume on.
Is the AI writing detection still running?
At some Australian universities and not at others, and several have switched it off since it launched. This is the part of Turnitin that has changed most.
| University | AI writing detection | What the university publishes |
|---|---|---|
| Curtin | Disabled from 1 January 2026 | Only the AI feature is off; text matching originality checks remain active |
| UQ | Made unavailable for Semester 2 2025 | UQ says it does not use AI detection tools, and is adopting more secure assessment |
| ANU | Not used for pursuing academic integrity matters, "at this time" | A library guide note from the tool's preview period, with no later decision published on that page |
| Melbourne | Available to staff | The report is not visible to students, and alone is not sufficient for an allegation |
| Sydney | May be used where unpermitted use is suspected | The score is considered alongside other relevant evidence |
Curtin announced that from 1 January 2026, the AI writing detection feature in Turnitin will be disabled, with effect across all campuses and study periods, gave fostering trust and clarity as the reason, and stated that only the Gen-AI detection feature is being disabled while the regular text-matching originality checks remain active. UQ's eLearning page records that the AI Writing Indicator was made unavailable at UQ for Semester 2 2025, the UQ Library's student hub states flatly that UQ does not use AI detection tools, and UQ's teaching guidance says the tool staff had was disabled in mid 2025 and that UQ is adopting more secure assessments to reduce its reliance on detection. ANU's position is older and still stands on the page. Its library guide carries a note written when Turnitin turned on a preview of the tool: the preview sat outside the University's regular contract, it is visible to staff but not students, and because it arrived part way through a teaching semester with no chance to evaluate it, it "will not be used at ANU for the purposes of pursuing academic integrity matters at this time". The guide has not published a later decision.
Melbourne kept it and published the caveats instead. Its advice for students on Turnitin and AI writing detection explains that the percentage represents text the tool calculates as 98 per cent or more likely to have been generated or paraphrased by AI, that detectors vary widely in how they work, and that a high score is a prompt for further investigation rather than a finding. The same page says the indicator and its report are not shown to students at all, so a number you cannot see may be sitting next to work you can. If a concern is raised, the instructor can share a PDF copy of the report with you, which is the moment to ask for it.
Turnitin's own release notes support the caution. From July 2024 the report stopped surfacing any score in the 1 to 19 per cent range, showing an asterisk instead, to avoid false positives. Reports handle up to 30,000 words of qualifying text, and submissions under 300 words do not meet the file requirements. And a December 2023 note states that an AI Writing score should not be used as the sole basis for adverse actions against a student.
Do not test your own work in a free detector
Do not run your own work through a free AI detector before you submit. Melbourne's page says free tools do not use the same methods, some are very inaccurate, and many exist to frighten you into buying a humaniser tool, which the University's detector often picks up anyway. It adds that uploading your work gives away your intellectual property, that the business model may involve training models on it or selling it to other students, and that this is a breach of academic integrity that could lead to a misconduct investigation. What happens to files you hand to a third party is a general problem, set out in student data privacy in Australia.
What happens from the first email to the outcome?
Melbourne publishes its process step by step, so it works as the worked example. Other universities differ in the names and the deadlines, not in the shape.
It starts before any allegation. Melbourne says being asked to discuss your work is not an allegation of academic misconduct, and that these discussions are informal and exploratory and may form part of the process that identifies or rules out a breach. You may be asked to explain what your work means, why you included particular ideas, or to share drafts. You are not obliged to take part in that stage.
If a case manager decides there is merit, a senior staff member has to approve it, and only then does an allegation notice go to your official University email address. The notice sets out the allegation, the possible outcomes, the support services and how to respond. You can respond in writing, be heard in person or by videoconference, or both. At the hearing the decision maker weighs the allegation, the evidence and your response, and decides on the balance of probabilities.
Melbourne then commits to a deadline. You receive a written outcome within five University business days, covering the decision, any penalties, the reasons for both, and your right to appeal. An appeal goes to the Academic Secretary within 20 working days of receiving the outcome. What the penalties themselves look like is set out in the ChatGPT rules at Australian universities.
What can you ask for?
More than most students realise, and asking is normal rather than aggressive.
- The evidence. Melbourne says you will be given the chance to review the evidence and submit your own before any decision is made. Where an AI report is involved, its advice page says your instructor can share a PDF copy with you.
- An independent adviser. Melbourne strongly recommends contacting the Student Union Advocacy Service before you respond, and describes it as free, confidential and independent of the University. Ask your own student union what its equivalent is called.
- A support person. You may bring one to the hearing, and you may attend, write, or do both.
- Reasons. The written outcome has to give the reasons for the decision and for the penalty, not just the result.
- An appeal. Within the published window, which at Melbourne is 20 working days.
- Time. Read the notice, then get advice before you reply. A fast reply written while you are upset is the one people regret.
Melbourne also says an AI writing detection report alone is not sufficient evidence for an allegation. If a score is the only thing on the table, that sentence is the one to point at. Bring what you have instead: the outline you wrote in week two, the reading notes with page numbers, the drafts with their timestamps, and the prompt log if a tool was permitted and you used one. None of it proves innocence on its own, and together it describes a process that a piece of generated text does not have.
How do you write so the report is boring?
By leaving a trail while you work, which costs nothing and settles the question in advance.
Write from your own notes rather than from the source text open in another window, because close paraphrase is what produces long unbroken highlights. Quote deliberately and cite every quote. Keep the reading notes, the outline and the dated drafts in one folder per assignment. If you used a generative tool where the task allowed it, keep the prompt log with them.
None of that is about beating a tool. It is the ordinary way of working that makes a matching report uninteresting and an explanation easy, and it is the same habit that makes revision possible later. The note-taking method behind it is in effective note-taking.
Where Notibo fits
Notibo does not check anything for plagiarism and has no view on your similarity score. It records a lecture on your computer, or takes an audio file you upload, and returns a transcript, structured notes and flashcards with spaced repetition. It also turns PDF, PPTX and DOCX files into notes and flashcards. It transcribes 89 languages, and the transcript is in the language spoken.
The first 14 days of Pro are free, with 240 minutes of recording and no card. When the trial ends, the free plan gives 30 minutes of recording a month. Pro is 9.99 euro a month or 88.99 euro a year, billed in euro with the local amount shown at checkout. A recording can run up to two hours, uploads are capped at 50 MB, and the transcript and notes export as PDF or Word. Files are stored in the EU on Supabase in Frankfurt, and transcription and the AI notes run on US processors under standard contractual clauses.
If you use notes it generated in assessed work, treat it like any other tool: check the AI statement on the task, acknowledge it if the task requires that, and write the argument yourself. Ask before you record anyone. More on how students use it is at students.