Short answer first: yes, and at several universities you can refuse. Waterloo prescribes a course outline statement saying text matching software may be used, and tells students who are worried about privacy to arrange an alternative in the first week of term, because submitted files sit on a server in California. The University of Toronto permits students to opt out and requires the instructor to check the work as rigorously another way. Simon Fraser gives instructors discretion to offer an alternative to a student with a principled objection. Toronto Metropolitan publishes an opt-out. What you are refusing is the upload, not the marking.
This article covers who uses the software, why the alternative exists in Canadian policy at all, what the report actually measures, where AI detection stands, and what a misconduct process looks at when a score is the only thing anyone has.
Which Canadian universities use text-matching software?
Enough of them that the more useful question is what your own course outline says. The pattern across the published policies is the same: the software is available to instructors, the instructor chooses whether to switch it on for a given assignment, and the outline has to say so. Where those rules came from, and what the surrounding privacy law actually gives you, is in student data privacy in Canada.
- Waterloo prescribes the wording. Waterloo's Office of Academic Integrity publishes the statement instructors must put on the course outline, and it includes the sentence that students' submissions are stored on a U.S. server and therefore students must be given an alternative if they are concerned about privacy or security. It names two: a scaffolded assignment or an annotated bibliography.
- Toronto prescribes a passage too. The University of Toronto's Centre for Teaching Support and Innovation gives instructors the sentence to include: normally, students will be required to submit their course essays to the University's plagiarism detection tool for a review of textual similarity and detection of possible plagiarism. U of T's academic integrity office repeats the obligation from the student's side: instructors who plan to use the tool must inform their students at the start of the course.
- SFU makes it a course requirement. Simon Fraser University tells instructors to state the intent to use Turnitin in the course outline, thereby making it a course requirement, and to explain that declining could mean being unable to complete the course.
- Toronto Metropolitan lists it as a tool. TMU's educational technology toolbox says the tool does have an opt-out policy and that students may ask their instructor about alternative assessment options.
The paragraph to look for in the syllabus
It is usually near the academic integrity section and it is short. Look for three things: whether text matching software may be used, where the submissions are stored, and whether an alternative is offered. If the first is there and the third is missing, that is the question to ask in week one, not in week ten.
Why did consent and alternatives come up in Canada at all?
Because the files cross a border. Waterloo states it in the plainest terms of any Canadian policy: submitted files are housed on a Turnitin server based in California, and these files are subject to the USA PATRIOT Act, 2001. That is the sentence that produced the opt-out. SFU says the same thing about its own obligations, noting that data submitted to Turnitin is stored and accessed on U.S. servers and that it updated its protocols so that practices continue to comply with British Columbia's Freedom of Information and Protection of Privacy Act.
Ontario institutions answer to their own version of the same statute. TMU's page says the submission data is retained within its learning system for two years, in case of potential appeals or FIPPA requests, which tells you two useful things at once: the record exists, and you can ask for it.
None of this gives you a right to keep your essay inside Canada. What it gives you is a right to be told and, at most Canadian universities that publish a policy, a right to ask for a different way of proving the work is yours. The general shape of your rights over data a university or an app holds about you, including how to make an access request, is covered in the privacy article linked above, and there is no point in repeating the law here.
Can you actually refuse, and what do you get instead?
Refusing is not the same as being excused. Every policy that offers an alternative replaces the upload with a different burden of proof, and the burden usually takes more of your time than the upload would have.
- Toronto. Students are permitted, under the University's conditions of use, to opt out of the plagiarism detection tool. The instructor must then find alternative arrangements to check their work as rigorously, and the examples given are an annotated bibliography, submitting rough work, or including source citations with call numbers.
- Waterloo. Students with privacy or security concerns may opt out and choose another alternative with the instructor, and the guidance says to contact the instructor as soon as possible and discuss an alternative within the first week of the term.
- SFU. If a student has a serious, principled objection to using Turnitin, the instructor has the discretion to offer an alternative process. SFU also notes that students may choose to use an anonymous identity to register with Turnitin, which is a smaller step than opting out.
- Toronto Metropolitan. The opt-out exists, and the page tells students to ask the instructor about alternative assessment options, "if available". That qualifier is doing work: the alternative is not guaranteed for every assignment.
The practical rule that comes out of all four is the same. Ask in week one, in writing, and ask for the alternative rather than for an exemption. An instructor who has to build an alternative in week eleven for one student will usually say no, and will be right to.
What does the software actually compare?
Turnitin's own guide is more modest about this than its reputation. It states flatly that Turnitin does not check for plagiarism in writing and that submissions are checked against its database, with matches highlighted for a human to review. The database, on the same page, is three things: web pages, current and archived; a repository of works students have submitted to Turnitin in the past; and a collection of periodicals, journals and publications.
Waterloo explains the arithmetic. The percentage represents the number of phrases the program finds in common with database content, divided by the total number of words in the file. So a 10 percent score means ten percent of the document matched sequences of phrases somewhere in the database, in one passage or scattered across several.
Two consequences follow, and both are common. Quoted text matches, because quotation marks are not a signal the matcher reads. And your own earlier draft matches, if it went into the same dropbox. Turnitin's guide gives that as a worked example: a student who submits drafts of the same paper can end with a final draft scoring 100 percent.
What is a similarity score not?
It is not a verdict, and no Canadian policy treats it as one. Waterloo's student guidance sets out the colour bands, blue at zero percent, green from one to 24, yellow 25 to 49, orange 50 to 74 and red 75 to 100, and then says the thing every student should read twice: there is no safe colour or percentage, and no percentage or colour in the report can fully evaluate whether text has been plagiarized.
| What the report shows | What it does not show |
|---|---|
| The share of text matching sources in the database | Whether the work is plagiarized |
| Which passages match, and a link to each source | Whether a match is a correctly cited quotation |
| Matches against web pages, journals and past student work | Matches against anything not in the database |
| Your own earlier draft, if it went into the same dropbox | Who wrote the text |
| A colour band from blue to red | A safe threshold |
Waterloo also points out the reverse case, which students rarely think about. Its guidance says Turnitin will identify direct quotations as matched text and that this will be reflected in the similarity score, but that as long as those quotations are properly cited, plagiarism has not occurred, and it tells you to go through the highlighted sections rather than focus on the percentage. Turnitin's guide says the same thing about its own database: even when a submission has quotation marks and references, the quoted text will show as a match. So the useful move when you get a report is to ignore the number and open the highlights.
Is AI writing detection switched on?
Not everywhere, and the trend in Canada is away from it. The University of Waterloo's Associate Vice-President Academic published the decision: the AI detection tool in Turnitin would no longer be available to University of Waterloo users as of September 2025. The reasons are listed on the page. Research shows the tools are unreliable. They are biased toward students whose first language is not English. Internal testing by the university's own instructional technology group found the advantages inconclusive, with the product flagging human written text as 100 percent generated by AI. The page's conclusion is that time and effort are better spent on education than on policing.
McGill wrote the same judgement into its rules instead of its practice. Article 25 of the Code of Student Conduct and Disciplinary Procedures forbids using generative AI in an assessment in a way the instructor has not authorized, and then adds a sentence that changes how any such case has to be built: AI-detection software is not sufficient, by itself, to substantiate an allegation under that article.
Turnitin says a version of this about its own product. Its guide states that the AI writing detection model may not always be accurate and should not be used as the sole basis for adverse actions against a student. The same page says the AI percentage is different from and independent of the similarity score, that AI highlights do not appear in the similarity report at all, and that scores above zero and below 20 percent are not surfaced because of false positives.
None of that makes unauthorized AI use safe. It means the case against you cannot be a percentage, and the case for you cannot be a percentage either. What your instructor is permitted to let you use is a separate question, and the ground rules are in AI for studying and in the assessment rules covered in how Canadian university exams work.
What does a misconduct process look at besides a score?
McGill publishes the clearest step-by-step version in the country, and it is worth reading before you ever need it. Under article 52 of the Code, an instructor with reasonable grounds to believe a student committed an academic offence refers the matter to the Disciplinary Officer of the faculty in which the course is offered, and the instructor may not, on their own authority, impose any sanction on the student. That single rule kills the idea that a high score can cost you a grade quietly.
What happens next is a process with dates attached. Under article 60, the Disciplinary Officer summons you to an interview by written notice. The interview takes place at least 5 and no more than 30 working days after the summons unless both sides agree otherwise. The notice has to describe the alleged offence, name the provision of the Code, and tell you that the relevant documentation may be consulted by you and your Advisor at least 5 days before the interview. You have the right to be accompanied by an Advisor, and the proceedings are confidential.
Article 61 adds the part most students do not expect: at the outset of the interview the officer must tell you that you are not obliged to answer any questions, and that any answers may become the basis for an immediate disposition or later testimony. Article 62 then requires a finding of founded or unfounded, and the Code's definitions set the standard of proof as the balance of probabilities, which it defines as proving on a preponderance of the evidence that a version of the facts is more likely true than not true.
Waterloo describes the same handover more briefly: the instructor evaluates the similarity report to assess whether the work is original, and any instances of suspected plagiarism are sent to the Associate Dean, Undergraduate or Graduate Studies, for review.
What to do the day you get the email
Read the notice for the article number and the date. Ask for the documentation the notice says you may consult, and ask for it in writing. Find out who can act as your Advisor, because most universities have a student advocacy service that does this all day. Then assemble the trail below before the meeting, in date order. Do not reply with an explanation until you have read what they are actually alleging.
What should you keep while you work?
The trail, not the finished file. A finished essay proves nothing about how it came to exist, and a dated set of drafts proves almost everything.
- Dated drafts. Write in something that keeps version history, or save a copy at the end of each session with the date in the filename. Three saved drafts a fortnight apart answer a question that a single file cannot.
- Notes with the source attached. Every quotation and every paraphrase carries the source and the page from the moment you write it down. The half hour after the reading is when this costs nothing and the week before the deadline is when it costs everything. The habit is the whole subject of effective note taking.
- The reading you did not use. An annotated bibliography is the alternative several Canadian universities offer instead of the upload, which tells you how much weight it carries as evidence.
- Submission receipts. Waterloo tells students to keep all email confirmation receipts until the course is over, as proof of submission. That is advice for every dropbox, not just Turnitin's.
- The brief and any change to it. Keep the assignment instructions as issued, plus any email or announcement that changed them, especially anything about what AI use is permitted.
None of that is defensive paranoia. It is the same material that makes the essay easier to write, which is why the students who keep it tend not to need it.
Where Notibo fits
Notibo is not a similarity checker and it does not detect AI writing. It sits earlier in the process: it records a lecture, or takes an audio file you upload, and returns a transcript, structured notes and flashcards with spaced repetition. It does the same with a PDF, PPTX or DOCX file. Where that helps here is in the trail. Notes that carry the lecture they came from, with a transcript behind them, are the kind of working record that answers a question about where a paragraph came from.
Transcription covers 89 languages, and the transcript comes back in the language spoken. Pro is free to try for 14 days with no card, and the trial includes 240 minutes of recording. Afterwards the free plan gives 30 minutes of recording a month. Pro is 9.99 euro a month or 88.99 euro a year, with the local amount shown at checkout. Uploads are capped at 50 MB. Files are stored in the EU, on Supabase in Frankfurt, and transcription and the AI notes run on US processors under standard contractual clauses, which is the same border question this article started with, and you are entitled to weigh it the same way. Recording a class needs the instructor's permission first. What the tool is for is in students.