Is using a paraphrasing or humanizing tool academic misconduct?

How universities define unauthorised assistance

The default is no, and the exception has to be written down. The University of Oxford's student guidance on generative AI says any use of AI in work submitted for assessment is unauthorised unless students are told otherwise in writing, in advance, for that specific assessment. Unauthorised use "constitutes cheating and plagiarism", with penalties up to failing the assessment and, in serious cases, expulsion. Where AI use is permitted, students must make a formal declaration in the format the assessment setter prescribes.

The University of Melbourne's student AI page states that use of generative AI tools in preparing assessed work must be acknowledged in all cases, and that unacknowledged use is academic misconduct. The subject coordinator decides what is permitted.

Where paraphrasers specifically fall

Whether these tools work against detectors is a separate question, covered in the humanizers and detectors thread. This thread is only about what the rules say.

Melbourne's advice on translation and editing tools treats spelling and grammar checkers, readability suggestions, paraphrasing, summarising, style rewriting and translation as one family. Using them to correct spelling, grammar or punctuation, or to make minor edits to work you completed yourself, "may be acceptable" if the coordinator permits it, you acknowledge it, and it does not change the meaning of your work. Submitting translated or rewritten content generated by a tool as if it were your own original work is misconduct.

Its test is useful for anybody: could you explain the ideas, arguments and wording in your own words if asked? It also warns that the outputs of these tools may be picked up by AI writing detection and that staff may ask you to discuss your work. If that happens, this thread on responding to a flag covers what to bring.

Two things follow. The origin of the draft matters more than the tool: a grammar pass over your own paragraph and a paraphrase pass over a chatbot's paragraph use the same software and are not the same act. And purpose matters: a rewrite whose only function is to lower a detector score is hard to describe as "editing" to an integrity panel.

What the style guides say about disclosure

The MLA Style Center's guidance on citing generative AI, first published in 2023 and revised in August 2025, says to cite an AI tool whenever you paraphrase, quote or incorporate content it created, and separately to acknowledge "all functional uses of the tool (like editing your prose or translating words)" in a note, in the text or another suitable place. For a student, that is the template: even a permitted editing pass gets a one-line acknowledgement.

Six common scenarios and how policies typically treat them

Scenario	Typically allowed?	What to disclose
Spelling and grammar check on your own draft	Usually, unless language itself is assessed	Tool name and what it did
Paraphrasing tool over your own draft	Only where editing tools are expressly permitted	Tool, passages affected, meaning checked
Translating your own draft from another language	Varies; often barred where English expression is assessed (see the bias thread)	Tool, source language, coordinator permission
AI generates the draft, you edit it	Only where generation is expressly permitted	Prompts, tool and version, what you changed
AI generates the draft, a humanizer rewrites it	No; treated as cheating and plagiarism	Nothing would make it compliant
AI for brainstorming or feedback only	Often permitted, still usually acknowledged	A note describing the functional use

What that means in practice

• Check the assessment brief, not the university homepage. Permission is set per assessment.
• Disclose functional use too. Editing and translating count, not just quoted content.
• Keep the draft that went in. The before-and-after is your evidence that the ideas were yours.
• Ask before, not after. A coordinator's written answer outweighs any general policy page.