Plagiarism and AI at Manchester: the viva voce, and what Cadmus records

The short answer

Manchester does not run AI detection. If your authorship is doubted, the answer is a conversation about your work — and a viva cannot, on its own, penalise you.

There is a second thing worth knowing, which almost nobody writes about: in some parts of the university, the submission platform records how long you spent producing the work.

The viva voce, and its limit

Where authorship is in question, Manchester’s procedure sends the case to a viva voce. Two sentences define what it can and cannot do:

“The staff conducting the viva voce will confirm that they accept that the student wrote the work in question; no further action will be taken. … If the staff conducting the viva voce still remain doubtful of the authorship of the work in question, or the student admits that it is not their work, then the case should be referred for disciplinary action. The viva voce, in itself, will not result in a penalty being applied; a penalty can only be applied by a disciplinary panel.”

That is a genuine protection, and it should change how you approach the meeting. The viva is where a doubt can simply end — “no further action will be taken” — and nothing worse than a referral can come out of it. So go, and go prepared, rather than treating it as a trap.

Preparing means being able to do what any author can: explain why you framed the question that way, what you read that changed your mind, which sources you rejected and why. Bring drafts and version history if you have them. Our guides on proving you wrote it yourself and what to do when you are accused set out the detail.

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.

What Cadmus records

Manchester’s procedure names a second platform alongside Turnitin, and the sentence deserves attention:

“References in this document to Turnitin should be read as to include alternative software which the University uses and identifies. For example, Cadmus is used in some areas of the University, and the data from its learner analytics may provide an indication of academic malpractice e.g. time spent on producing the work.”

This is behavioural evidence, not textual. It does not look at your prose; it looks at your working pattern. A dissertation that appears in the platform in a single short session looks different from one written over weeks — and at Manchester that difference is explicitly named as a possible indicator.

The honest implication is not sinister, and it cuts in your favour more often than against you: if you actually did the work in the platform, that record is evidence for you. The students it disadvantages are those who draft everything elsewhere and paste a finished document in at the end — which is a very common and entirely innocent habit.

If your course uses Cadmus, doing at least some of your drafting inside it is a cheap insurance policy.

Software that claims to detect AI

The statement is inside the Academic Malpractice Procedure itself — approved by the Student Conduct and Discipline Committee, not a guidance page:

“The University does not currently use software which claims to detect Generative AI; common indicators of malpractice, and information sources, should be reviewed as with other forms of suspected malpractice.”

Read the wording carefully, because it is doing more work than it looks. Manchester did not write “we have disabled the AI indicator”. It wrote that it does not use software which claims to detect generative AI. That is a statement about the claim, not just the configuration.

It entered the procedure in version 5 of November 2023 — the amendment history records that version as the one that “include[d] reference to Generative AI” — and survives into the current version 6 of March 2025.

On similarity, Manchester uses Turnitin and is clear about its weight: “Turnitin alone does not confirm the presence of malpractice; an academic judgement must be made to reach this conclusion.” On what a similarity percentage actually measures, see what a Turnitin score actually means; on the reliability question generally, how accurate AI detectors really are. The same limit applies to our own tool: our published error rates are measured on English and German corpora, and no detector produces proof.

Manchester discourages pre-checking

We should quote this even though it is inconvenient for us, because you need to know it before you act:

“Students are discouraged from using alternative software in an attempt to identify malpractice prior to the submission of an assessment; alternative software can be less comprehensive than Turnitin and return inaccurate results.”

That advice covers third-party checkers generally — ours included. If you study at Manchester, take it at face value. No university-provided student self-check area is documented either, so the practical answer is to write carefully, cite as you go, and use the viva route if a question ever arises.

There is a further reason to be careful about where you upload a draft, which Manchester does not spell out but which is real everywhere: some checking services store what you give them and compare later submissions against it, so your own draft can come back as a match against you. Our own check does not add your text to any third-party comparison database — but at Manchester, the university’s advice is the one that governs.

AI sits inside plagiarism

Manchester made a structural choice here that differs from Nottingham’s or Sheffield’s. There is no separate AI offence. AI is written into the definition of plagiarism:

“Plagiarism is the act of using ideas, words, or creations from either humans or digital systems such as Generative Artificial intelligence (AI) without proper attribution or permission and presenting them, either intentionally or unwittingly, as one’s own work. Plagiarism encompasses a range of practices beyond just verbatim (word for word) copying. It also includes instances of close paraphrasing, minimal adaptation, and other actions that involve reproducing the work of another source in a way that means the assessed work lacks appropriate originality or proper referencing.”

“Close paraphrasing, minimal adaptation” is the part to note. Changing words while keeping someone else’s structure is inside the definition, which is exactly what an automated rewriter produces.

The umbrella definition covers the unintentional case too: malpractice is “any activity – intentional or otherwise – that is likely to undermine the integrity essential to scholarship”, and “whether intended or not, all incidents of academic malpractice will be treated seriously”.

Self-plagiarism is also caught: reusing your own earlier submitted work without citation “is considered to be plagiarism because it misrepresents the novelty of the current work.”

Where AI use is permitted, ordinary referencing is the mechanism — our AI disclosure statement guide covers the shape of a good acknowledgement.

Beyond basic proofreading

Manchester’s contract cheating definition contains a useful boundary that most policies leave implicit:

“Contract cheating is the commissioning of a piece of work by a third party, beyond basic proofreading. […] This may also include the use of crowdsourcing, where a student obtains content from, or allows editing by, others and fails to acknowledge the contribution.”

Two things follow. Basic proofreading is expressly outside the offence — so having someone catch your typos is fine. And the second sentence catches something people rarely think about: posting a paragraph in a group chat and letting someone rewrite it is “allowing editing by others”, and failing to acknowledge it is the offence.

Collusion is defined broadly too, including “sharing of work, ideas or plans by social media or other electronic communication means”.

How far back it can look

A limit worth knowing, because at some universities there is none:

“The detection of academic malpractice will not normally lead to the review of a student’s previous assessments, unless there is a reasonable suspicion that they may contain malpractice. If this is the case […] the AUO may review previous pieces assessments, but this will not normally extend to assessments outside the student’s current level of study.”

So one finding does not normally trigger an audit of everything you have ever submitted, and a master’s case does not normally reach back into your undergraduate work.

Two documentary notes we would rather flag than hide. The student-facing guidance is version 2.6 of October 2023 and carries “Next review date: August 2024” — which has passed. And the Academic Malpractice Procedure’s own control box says “Version: Unknown (at least 4)” while its amendment history runs to version 6. Neither undermines the rules; both are reasons to check with your school rather than assume a page is current.

Before you submit

• Draft inside Cadmus if your course uses it. The working record is more likely to help you than hurt you.
• Do not use an outside plagiarism checker. Manchester discourages it directly.
• Attribute AI use through normal referencing. It is inside the plagiarism definition, so an unacknowledged use is a plagiarism finding.
• Watch close paraphrasing.Keeping someone else’s structure while changing words is named in the definition.
• If you are called to a viva, attend. It cannot impose a penalty, and it can end the matter outright.

If you want a read on your own text before you hand it in, our AI checkis free up to 1,500 words, with no account and no name — but note Manchester’s own advice above, and follow it while you are its student.

Sources

• University of Manchester: Academic Malpractice Procedure — current text version 6, March 2025; approved by the Student Conduct and Discipline Committee.
• University of Manchester: Guidance to students on plagiarism and other forms of academic malpractice — v2.6, October 2023, approved by the Teaching and Learning Strategy Group; stated next review date August 2024.

This page is orientation, not legal advice. What binds you is Regulation XVII and the Academic Malpractice Procedure.