Can you use ChatGPT at a New Zealand university?

Short answer first: you may use generative AI only where the assessment allows it, and all eight universities leave that call to the course or paper. Auckland sorts assessments into Lane 1, where AI is restricted, and Lane 2, where it is not, with every course using the labels from 2027. AUT uses Channel 1 and Channel 2, Massey gives each task one of four AI levels, and Lincoln prints a No, Partial or Full Usage statement and treats silence as a ban. Acknowledge any use; Waikato says always. All eight state that unauthorised use breaches academic integrity, with penalties up to exclusion.

The comparison below draws on the AI policies, student guidance and misconduct rules the eight universities publish, as they stood in September 2026. How the rules meet exams and revision is in the exam preparation guide.

Who decides whether you can use AI in an assessment?

Not the university as a whole. Each one writes a framework and hands the actual decision to whoever runs the course or paper, often assessment by assessment, so two assignments due in the same week can carry opposite rules. The instruction that binds you is the one in your own outline or brief.

• Victoria puts it with the course coordinator. Clause 5.6 of Te Herenga Waka Victoria University of Wellington's Generative Artificial Intelligence Policy has course coordinators set their approach in the course outline or the documents for each assessment item, allowing or barring AI in whole or in part.
• Canterbury wants it in writing. Clause 3.2 of its Assessment Policy and Principles makes the course coordinator state whether AI is permitted, to what extent, with which tools and how to acknowledge it.
• Otago sets a duty for every assessment. Its policy on generative AI in learning and teaching requires staff to give written guidance on acceptable use for each assessment, naming the permitted tools, and to warn that unacceptable use is misconduct.
• Waikato uses the paper outline. Its Student Assessment Handbook page on generative AI points to the Paper Outline Tool in Moodle and tells students to check with the paper convener or lecturer when unsure.
• Auckland uses the course outline and Canvas. Its guidelines on permitted use of software say the Course Director clarifies what a course allows, and software used outside that is not considered acceptable.

So read two documents before you start: the outline and the brief for the task. If they disagree or say nothing, email the coordinator or convener and keep the reply.

How do the lanes, channels and levels compare?

Each university has built its own scale, and they do not line up neatly. Auckland and AUT split assessments in two, Massey and Lincoln grade the permitted use in levels, and Canterbury describes the range in plain words. The table adds what each says applies when the instructions are silent.

• Auckland: two lanes. Its advice for students places interactive orals, practical work and invigilated tests in Lane 1, where AI is restricted unless a tool is expressly allowed, and essays, reports, case studies and presentations in Lane 2, where use is not restricted. Every course will label its assessments by lane from 2027.
• AUT: two channels. AUT adapted the University of Sydney's two-lane model into Channel 1 and Channel 2: Channel 1 tasks are invigilated at a set time and exclude AI tools, and Channel 2 tasks run in any setting with AI permitted. AUT's student page on using AI says to start with Canvas, ask the lecturer or tutor when unsure, and be open about how you used it.
• Massey: four levels. Its AI usage and detection page lists No AI, AI planning, AI collaboration and AI exploration, allocated per assessment, and bars AI from secured assessments such as practicals, invigilated tests and exams.
• Lincoln: three usage levels. Its student guidelines have every assessment carry a GenAI Usage statement: No Usage, Partial Usage for idea generation, refinement or content creation, or Full Usage.
• Canterbury: four possibilities. Its Gen-AI FAQs for students say use may be fully excluded, restricted to certain uses or tools, completely unrestricted or even required, depending on the assessment.

What if the instructions say nothing?

Lincoln's Gen AI and assessments page is the bluntest: with no advice on the course page or in the instructions, assume AI is not allowed, which is also how its discipline process reads a case. Canterbury tells you to contact the course coordinator before assuming. Waikato bars AI-written or AI-rewritten answers unless the instructions permit them, and Massey's Generative Artificial Intelligence in Assessment Policy limits use to what the assessment criteria explicitly allow.

Which uses are usually fine, and which cross the line?

Victoria sorts common uses into typically good, questionable and typically bad, and Massey and Lincoln add examples of their own. None of these lists overrides your course.

• Study aids are typically good at Victoria. Its page on student use of artificial intelligence lists consolidating notes written in your own words, generating questions to test yourself, checking grammar and making flashcards from your notes.
• Planning and summaries are questionable there. Outlining an argument, writing a reference list, summarising articles and summarising lectures sit in the middle: planning may be part of the task, AI gets reference lists wrong, and a summarised lecture can miss what you were meant to learn.
• Doing the work is typically bad. Writing sections of an assignment, answering test questions, inventing references and uploading material you do not own, such as slides or recorded classes, fall on the wrong side.
• Massey ties uses to the level. At the planning level you may organise your thoughts or summarise your notes for revision; the collaboration level adds grammar checks on text you wrote yourself.
• Lincoln rules out translation. Its advice says never to translate an assessment from your first language with AI, and warns that translation software will often get work flagged for AI use.
• Privacy is a separate test. Waikato warns that uploading work to non-institutional AI platforms carries privacy and ethical risks, because such tools often lack safeguards aligned with the Privacy Act; what that Act gives you is in the student data privacy guide.

Victoria's page on academic misconduct offers a test for when the lists run out: if you would not ask another person to do it, you probably should not ask AI either. Work that builds the skill being assessed belongs to you. Help with understanding or checking your own draft is more often allowed. Ways to revise with AI that stay on the right side of these lines are in the guide to AI for studying.

Do you have to say you used AI?

Wherever a course allows AI, expect to declare it. The form differs, from one sentence to a full appendix, and some universities want material you may not think to save, such as prompts. Read the acknowledgement rule before you open the tool.

• Waikato: always. Its assessment handbook page on academic integrity tells students to always acknowledge generative AI used in assessment tasks, and says the lecturer can advise how.
• Victoria: follow citation standards. Clause 5.11 of its policy requires students to acknowledge any use in line with appropriate citation standards and library guidance, and its student page suggests keeping a log of the tools and prompts you use.
• Canterbury: describe the use. The FAQ says you will be required to disclose use with a description of how you used it, on a cover page, in a section of the assessment or, in some courses, on a declaration form.
• Massey: an appendix. The AI usage page asks for an appendix with every tool and prompt used, the AI output, and where and how you changed it; the policy lets a course also require all prompts and a full version history.
• Lincoln: tool, method, prompts and time. The guidelines set the minimum as the tool, how it was used including the prompts, and when, and a No Usage assessment needs a statement that no AI was used.
• Auckland: when the course asks. The software guidelines say you may be required to declare a tool, with the Digital Course Outline and Canvas setting the format.

The safe habit is the strictest one: record every tool, prompt and date as you go, even where the course only wants a sentence.

What counts as misconduct with AI?

Auckland's guide for staff puts it plainly: using AI is not, "in and of itself", misconduct. The breach lies in using it where it is barred, hiding it where disclosure is required, or passing off its output as your own thinking. The universities also separate carelessness from deception, and that line decides how serious the outcome is.

• Auckland grades the problem. Its process for AI misuse in Lane 2 assessment treats generic AI-assisted work as a marking issue, failure to disclose AI use when required as poor academic practice, and knowingly submitting fabricated or hallucinated references as misconduct.
• Canterbury names it. Its 2024 misconduct procedures guide for students lists improper use of AI tools, meaning use that is not expressly permitted or a failure to acknowledge it as instructed, among its examples of academic misconduct.
• Massey sorts by size. Its page on breaches of academic integrity treats minor cases of unauthorised AI use as a minor breach and other unauthorised use as misconduct.
• Lincoln calls it ghost-writing. Its Academic Integrity Policy and Procedure includes unauthorised use of AI programmes in its definition of ghost-writing, and any such case goes directly to the Proctor.
• Victoria does not need intent. Its Student Academic Integrity Regulation counts a breach of its expectations, including the rules on generative AI, as academic misconduct whether inadvertent or deliberate.
• AUT calls it dishonesty. Its Academic Integrity Procedures and Guidelines list dishonestly representing AI-generated content as your own work among the types of academic dishonesty, and require clear instructions on use and acknowledgement wherever a course permits AI.

A missing acknowledgement can therefore be a breach even when the work is honest. The trap that belongs to AI is the invented reference, so check every reference against the real source before you submit.

What penalties do the universities publish?

All eight publish a ladder of outcomes. The bottom rung is educational and the top one ends a degree. Where a case lands depends on intent, the size of the breach, your stage of study and any earlier finding.

• Auckland limits what a department can do. For minor misconduct its Student Academic Conduct Statute allows a mark cut down to zero, a final grade lowered by up to three grade steps, or a reprimand; heavier cases go to the Discipline Committee.
• Otago adds a fine at Level Two. Its Student Academic Misconduct Procedures allow a zero, a failed paper or, where no direct academic advantage was possible, a fine of up to 500 New Zealand dollars or up to 40 hours of community work.
• Lincoln starts at zero for dishonesty. Its procedure calls a zero the most suitable penalty once dishonest intent is admitted or found.
• Victoria's top outcomes sit in its statute. Its Student Conduct Statute lists a cancelled mark or a failed course for academic misconduct, and for serious misconduct exclusion from some or all courses for up to two years.

How is a suspected AI case handled?

The case starts with the person who marks or runs your course and moves up only if it is serious. What differs is who holds the file first. The steps from the first email to an appeal, and what you may ask for on the way, are in the Turnitin guide.

• Massey starts with a conversation. Its AI usage page says an invitation to explain your work is not an allegation; the coordinator can then treat a problem as poor practice or refer it to an Academic Integrity Officer.
• Canterbury starts with an officer. Under its Academic Misconduct Regulations an Academic Integrity Officer investigates first and must give you a reasonable opportunity to respond, then can deny credit or refer the case to the Proctor.
• Lincoln skips a step. Course Examiners handle most first offences themselves, but cases of unauthorised AI use go straight to the Proctor.
• Otago splits by level. A Head of Department deals with Level One cases, and the Pro-Vice-Chancellor of the division with Levels Two and Three.
• AUT starts with an officer. Its procedures send every suspected breach to a faculty Academic Integrity Officer, who resolves minor cases and refers major or repeat ones to the faculty disciplinary committee.

Keep a record while you work

The strongest answer in any of these processes is evidence that you made the work. Canterbury's FAQ tells an accused student to collect earlier drafts, workings and planning notes before the meeting, and Massey's policy lets a course require a full version history. Write in a tool that keeps versions and export any AI chat you declared.

What should an AI acknowledgement say?

The universities ask for the same facts in different wrappers: which tool, what for, what you typed and what you did with the answer. Canterbury publishes copy-and-paste examples, Lincoln adds when you used it, and Massey adds the output itself. Build one record that holds all of it, then trim it to the form your course asks for.

1. Check the level or lane on the assessment, and whether the course names a format, a cover page or a form.
2. Log as you work: the tool, the date, each prompt and a copy of the output you used.
3. Write the statement from the log, using the fields below.
4. Attach an appendix or form where the course asks for one, such as Massey's record of prompts and outputs or a Canterbury declaration form.
5. Cite any AI output you quote or paraphrase in the referencing style the course uses.

A statement you can adapt

Canterbury's acknowledgement examples follow an order that works almost anywhere: the tool and its link, what you used it for, the prompts, and how the output was used. Add Lincoln's date and the statement covers the facts the others ask for; Massey also wants the output itself in an appendix. Laid out as fields, it reads like this.

• Tool: the name and link, for example ChatGPT and the address you used.
• Date: when you used it.
• Purpose: for example, generating practice questions or checking the grammar of a draft you wrote.
• Prompts: the exact wording, or a note pointing to the appendix.
• Use of output: what you kept, changed or discarded, and where it appears.

If you used no AI

Lincoln's No Usage statement asks students to acknowledge that no GenAI tool was used for any aspect of the assessment. The other universities read for this guide ask for a statement only when AI was used, but where a form has a box for no use, tick it.

Where Notibo fits

Notibo is an AI tool, so everything above applies to it. It records a lecture on a computer or takes an audio file you upload and returns a transcript and structured notes, and it turns PDF, PPTX and DOCX files into notes and flashcards. You get Pro free for 14 days, with 240 minutes of audio, and no card is asked for. After those 14 days the free plan covers 30 minutes of audio a month, with a transcript and notes. Flashcards with spaced repetition from a recording come with Pro, at 9.99 euro a month or 88.99 euro a year.

Use it on material you are allowed to put into an outside AI tool. Your own notes are the clearest case, and Lincoln's guidelines tell students to add only notes they created themselves. Slides and recordings are different: Canterbury wants the course coordinator's permission, Victoria lists recorded classes as a bad upload, and Lincoln bars course material, written or oral. Record a lecture only with the lecturer's permission.

What Notibo produces is revision material, not your work, so never paste it into an assessment. Files are stored in the EU, and transcription and the AI notes run on US processors named in the privacy policy.