Plagiarism and AI at McGill: what the Code says about detectors

The short answer

McGill has put the limits of AI detection into its Code of Student Conduct — not into a guidance page, not into an FAQ, but into the binding rulebook approved by Senate.

Several universities say detectors are unreliable. McGill is the only one in this series that made it a provision of the Code itself.

The sentence, and where it sits

Article 25(b), in full:

“No Student may use generative AI in the context of an Assessment in a manner not authorized by the relevant instructor or supervisor, or in a manner that violates any requirements regarding the use of AI in the context of a given course. AI-detection software is not sufficient, by itself, to substantiate an allegation under this article of the Code.”

Notice the structure. The first sentence creates the offence; the second limits how it may be proved. Both are in the same article, and both bind.

The distinction that matters is wherethis appears. Elsewhere in this series — UCL, Toronto, Penn State, Sheffield, Manchester — similar positions live in student guidance, a teaching centre’s advice, or an FAQ, and their status is always slightly soft. McGill’s sits in the Code of Student Conduct and Disciplinary Procedures, approved by Senate, in a version effective 12 November 2025 and confirmed by the Board of Governors on 11 December 2025.

The practical consequence, if you are ever facing an allegation, is that this is not a submission about best practice. It is a provision of the instrument under which the allegation is being brought, and it says a detector result on its own cannot substantiate it. Something else has to be there.

We hold ourselves to the same standard we would apply to anyone else’s tool: our published error rates are measured on English and German corpora, and no detector — ours included — produces proof. See how accurate AI detectors really are and what a Turnitin score actually means.

What does end a case, at McGill as everywhere, is your own record of working: drafts, dated version history, notes. See how to prove you wrote it yourself and what to do when you are accused.

If a detector score is what started this, our guide to an AI detector flagging your thesis sets out what that number does and does not establish on its own.

Two articles, two different rules

McGill catches AI in two places, and it is worth knowing both because they are triggered differently.

Article 25(b) — plagiarism. The offence is using AI “in a manner not authorized by the relevant instructor or supervisor”. The test is the absence of permission.

Article 26(f) — cheating. The offence is to “use any tool to complete an Assessment, including generative artificial intelligence, that has been disallowed in the course outline or any written instructions for the Assessment.” The test is the presence of a prohibition.

In practice that closes the gap from both sides: silence does not help you under 25(b), because you needed authorisation; and an explicit ban catches you under 26(f). The safe course is the same as everywhere — find out what is permitted, in writing.

The rest of article 25 covers the familiar ground: no student “shall represent work not authored by themselves, published or unpublished as their own in any writing, such as an essay, thesis, research report, project or assignment”. Note that thesis is named, so this reaches graduate work directly.

Your instructor sets the rule

McGill publishes no university-wide AI declaration form and no required wording. The Code deliberately devolves the question: what matters is whether the use was authorised by your instructor or supervisor, or disallowed in the course outline.

So the practical steps are short. Read the course outline. If it is silent on AI, ask — by email, so the answer exists in writing. Keep that reply with your drafts. Under both articles, whether you had permission is the whole question, and an email settles it.

If you want to be transparent about a permitted use anyway, our AI disclosure statement guide covers what a good acknowledgement contains — but follow your course’s form if it has one.

The clause about sharing your work

Article 25(c) catches something students rarely think of as an offence at all, and it puts the risk on the generous party:

“No Student may, without permission of the relevant instructor, Knowingly share their work produced for an Assessment with another Student who could submit that work in part or whole as their own.”

Sending your finished essay to a friend “just so they can see the structure” is the classic version of this. The clause bites on the sharing, not on whether the other person actually used it — the test is that they could submit it.

The same logic applies to uploading finished work to study-sharing sites. It is worth being careful where your own writing ends up, for this reason and because some services retain what you give them. For what it is worth, our own check does not add your text to any third-party comparison database.

What McGill does not publish

Four things we looked for and could not establish, stated as absences rather than filled in:

• Any named similarity product. No text-matching tool is named in the Code or on the Dean of Students academic-integrity pages, and the IT and teaching-support pages we probed returned 404.
• Whether an AI-writing indicator is enabled anywhere. McGill publishes no on/off statement. What it publishes instead is the evidentiary limit quoted above, which is arguably more useful to you than a configuration fact would be.
• Any opt-out or data-residency statement. There is no published alternative-submission route and no statement about where submitted work is stored — a contrast with Toronto and Waterloo, which both publish one.
• Any student self-check route, and any annual case counts.

One structural note worth knowing: McGill’s Code is under review in the 2026-27 academic year, so the version described here may be superseded. Check the Secretariat’s copy rather than relying on any summary, including this one, if something turns on the exact wording.

Before you submit

• Get AI permission in writing. Both articles turn on authorisation, and an email is the whole defence.
• Read the course outline for a disallowance. Article 26(f) bites on what the outline forbids.
• Do not share finished work with classmateswithout your instructor’s permission — the sharer is caught.
• Keep drafts and version history. A detector result cannot substantiate an allegation alone, and your working record is what fills the gap in your favour.

If you want a read on your own text before you hand it in, our AI check is free up to 1,500 words, with no account and no name.

Sources

• McGill University: Code of Student Conduct and Disciplinary Procedures — approved by Senate; initial approval 13 May 1981; current version effective 12 November 2025, confirmed by the Board of Governors 11 December 2025; next review 2026-27. Articles 25 and 26.
• McGill University: Academic integrity — Dean of Students.

This page is orientation, not legal advice. What binds you is your course outline and the Code of Student Conduct and Disciplinary Procedures.