Academic Misconduct Policies: How Different Universities Define Plagiarism

Policy snapshot: July 2026. University-wide rules may be supplemented by faculty, department, course, and assessment instructions.

Universities broadly agree that students must not present another person’s work or ideas as their own. Yet their academic misconduct policies do not use identical definitions. They differ in how they treat intention, previous work, collaboration, artificial intelligence, non-text materials, and weak academic practice.

These differences matter because plagiarism is not limited to copying an entire essay. It may involve an unattributed idea, a close paraphrase, a section of code, a figure, a dataset, a translation, or work produced with unauthorized assistance. Students therefore need to read the rules that apply to the exact assessment rather than relying on a general understanding of plagiarism.

The Shared Core of University Definitions

Most definitions contain three basic elements. A student uses material originating elsewhere, includes it in assessed work, and fails to acknowledge the source properly. The material may consist of words, ideas, analysis, data, images, code, or another form of intellectual or creative work.

Academic misconduct is usually a wider category than plagiarism. It may also include cheating, falsification, fabrication, collusion, contract cheating, impersonation, unauthorized collaboration, and prohibited technology use. A case can therefore violate academic integrity even when no matching source text exists.

Definitions Compared

This table is a summary, not a substitute for the complete policies. The applicable rule may also depend on the student’s program and the instructions attached to a particular task.

Harvard: Ideas and Language Require Credit

Harvard’s source guidance treats it as plagiarism to take an idea or language from another source without adequate credit. The source may be a published author, another student, a website, or another person. Its guidance also stresses that plagiarism may be intentional or accidental.

This approach makes attribution central. Exact wording requires quotation and citation, while paraphrased material must also be acknowledged. Harvard additionally warns that students must identify material derived from their own earlier writing when reuse rules require it.

Oxford: Broad Sources, Consent, and AI

Oxford defines plagiarism as presenting work or ideas from another source as one’s own without full acknowledgement. The definition covers published and unpublished material in manuscript, printed, and electronic forms. Permission from the original author does not remove the need for attribution.

Oxford’s current guidance also includes material generated wholly or partly through artificial intelligence when AI use has not received authorization for the assessment. Its materials distinguish intentional, reckless, and unintentional plagiarism, although intention may affect how a case is classified and handled.

Stanford: The Importance of Permitted Aid

Stanford’s Honor Code is framed around unpermitted academic aid. Students must neither give nor accept assistance that has not been authorized for work used in evaluation.

This model makes course instructions especially important. Collaboration, editing, tutoring, software, or AI assistance may be acceptable in one assignment and prohibited in another. Proper citation does not automatically solve the problem when the assistance itself was not permitted.

MIT: Plagiarism Beyond Written Sentences

MIT’s academic integrity materials describe plagiarism as using another person’s words, ideas, research, data, or figures without appropriate credit. Plagiarism appears within a broader framework that also includes cheating and unauthorized collaboration.

The inclusion of data and figures is important for technical work. Students can misrepresent authorship through copied diagrams, experimental results, code, models, or analytical claims even when a conventional essay contains little matching prose.

UCL: Separate Categories for Different Conduct

UCL defines plagiarism as representing another person’s work or ideas as the student’s own without appropriate acknowledgement. Its procedure separately defines self-plagiarism as reproducing or resubmitting previously assessed work and collusion as unauthorized collaboration by two or more students.

UCL also connects generative AI with assessment-specific permission. Using AI beyond what the assessment brief allows can fall within academic misconduct. Its framework can distinguish poor academic practice from a formal misconduct finding, allowing the nature and extent of the problem to be considered.

Cambridge: Intent to Deceive Is Not Essential

Cambridge defines plagiarism as using someone else’s ideas, words, data, or other material without acknowledgement. Its rules expressly state that a student may commit plagiarism regardless of an intention to deceive.

Cambridge also identifies self-plagiarism, collusion, contract cheating, and fabrication as forms of academic misconduct. This shows why an accidental attribution failure may still breach the rules even though intent remains relevant when evidence and outcomes are considered.

Melbourne: Text, Code, Design, and Creative Work

The University of Melbourne applies academic integrity principles across many forms of work. Its guidance addresses writing, interpretations, code, software, designs, music, sounds, images, and photographs.

It also treats unauthorized reuse of previously submitted work as an integrity concern. Generative AI rules can vary by subject and assessment, so students are directed to check the instructions set by the subject coordinator rather than assume that one university-wide permission applies everywhere.

Toronto: Plagiarism Within a Formal Conduct Code

The University of Toronto’s current Code of Behaviour on Academic Matters took effect on July 1, 2025. The code places plagiarism within a broader system governing academic offences and procedures.

University guidance discusses plagiarism alongside conduct such as submitting the same work more than once, purchasing work, cheating, and impersonation. The structure illustrates that institutions may regulate duplicate submission or false authorship as related but distinct offences.

Does Intent Matter?

Intent is one of the most important policy differences. Harvard guidance includes intentional and unintentional plagiarism. Cambridge states that intent to deceive is not required. Oxford recognizes intentional, reckless, and unintentional forms.

This does not mean that intent is irrelevant. It may influence whether conduct is treated as poor practice or misconduct, how serious the case appears, and which response is proportionate. However, “I did not mean to plagiarize” may not remove the underlying failure to acknowledge a source.

Self-Plagiarism and Duplicate Submission

Self-plagiarism does not involve taking authorship from someone else. The problem is presenting previously assessed work as new work completed for a new requirement.

UCL and Cambridge expressly define unauthorized reuse of earlier work, while Melbourne also addresses reuse without permission or acknowledgement. Toronto may identify submitting the same work twice as a related offence. Reuse can be acceptable when an instructor authorizes it, a draft forms an approved part of the same assessment, or the earlier work is disclosed and cited as required.

Plagiarism, Collusion, and Contract Cheating

Collusion concerns unauthorized joint work. Students may cite outside sources correctly yet still breach the rules by co-writing an individual assignment or sharing completed answers.

Contract cheating involves submitting work produced by another person or service. The commissioned work may be original and produce no text match, but the student still misrepresents authorship. This is why similarity software cannot detect every form of academic misconduct.

How Generative AI Changes the Comparison

Universities use several models for AI. Oxford includes unauthorized AI-generated material within its plagiarism guidance. UCL connects misconduct to use that exceeds the assessment brief. Melbourne emphasizes subject-specific permission, while Stanford’s unpermitted-aid framework asks whether the assistance was authorized.

The same activity may therefore be allowed in one course and prohibited in another. Students should check whether AI may be used for brainstorming, outlining, translation, editing, coding, or drafting and whether its use must be disclosed. They remain responsible for invented references, inaccurate claims, and material that does not represent their own assessed learning.

Why Citation Is Not Always Enough

A citation may identify a source without making every use acceptable. Problems remain when exact words lack quotation marks, a paraphrase follows the original too closely, another person produced the analysis, collaboration was unauthorized, or an old assignment was reused without permission.

Academic policies therefore regulate more than attribution. They also protect independent authorship, permitted assistance, and the learning outcomes that an assessment is designed to measure.

Similarity Reports Are Evidence, Not Verdicts

Text-matching software identifies similarities between documents. It does not determine intent or decide whether plagiarism occurred.

Legitimate quotations, references, templates, and standard phrases can create matches. Contract cheating, translated copying, copied ideas, and some unauthorized AI use may produce a low similarity score. Academic judgment, source review, drafts, assignment rules, and the student’s explanation remain important.

Definitions and Procedures Are Different

A definition explains which conduct is prohibited. A procedure explains who investigates, what evidence is considered, how the student can respond, who decides the case, and whether an appeal is available.

Universities should therefore not be ranked as simply “strict” or “lenient” based on the maximum possible penalty. Outcomes often depend on the scale of the problem, the assessment’s value, prior instruction, previous cases, intention, and institutional procedure.

A Practical Checklist for Students

• Read the university policy, course syllabus, and assessment brief.
• Cite borrowed ideas as well as exact words.
• Use quotation marks or block formatting for copied wording.
• Credit figures, datasets, images, code, and translations.
• Confirm what collaboration and outside help are permitted.
• Ask before reusing previously submitted work.
• Check whether AI use is allowed and must be disclosed.
• Verify that every listed source exists and was actually consulted.
• Keep notes and drafts that show how the work developed.
• Request clarification before submission when a rule is unclear.

Conclusion

Universities share a central principle: students must acknowledge material that is not their own and must represent authorship honestly. The differences appear in how broadly material is defined, whether intent is required, how previous work is treated, and whether AI use is classified as plagiarism or unauthorized assistance.

Harvard emphasizes ideas and language, Oxford covers broad source types and unauthorized AI material, Stanford focuses on permitted aid, and MIT explicitly includes data and figures. UCL separates plagiarism, self-plagiarism, and collusion; Cambridge does not require intent to deceive; Melbourne applies the rules across technical and creative work; and Toronto places plagiarism within a wider conduct code.

The safest approach is to treat the university policy as a starting point, not the only rule. Students should also read faculty guidance, the course syllabus, and the exact assessment instructions before using sources, collaborators, previous work, or AI tools.