Can you use ChatGPT at an Irish university?

Short answer first: there is no national ban and no national permission. Trinity's College Statement on AI and GenAI, issued on 4 June 2026, hands the detail to schools and degree programmes. UCD requires every module descriptor to say whether students may use AI and under what conditions. University of Galway says you must not present generative AI output as your own work, and UCC's academic integrity policy says the same in its own words. So the rule that binds you is the one on your module page, and the default answer everywhere is that undeclared use of it in submitted work is misconduct.

What follows is what five named Irish universities actually publish, and what the national guidance behind them says. Your own institution may word it differently, so treat this as the shape of the system and read your handbook for the detail.

Who actually makes the rule for your assignment?

Not the university, in most cases. The policies set the floor, and the module sets the rule you will be marked against. That is why two modules in the same degree can give you opposite answers in the same week, and why "my other lecturer said it was fine" is not a defence.

• UCD puts it in the module descriptor. UCD's guidelines say a clear statement on the use of AI within a module must be included within the module descriptor, confirming whether student use is permissible and under what conditions.
• Trinity pushes it down to schools. The College Statement says regulation can range from how usage is acknowledged or cited in submissions to outright prohibition of GenAI in producing them, and that these rules belong in the discipline or degree programme handbook.
• Galway leaves room for the instructor. Its FAQs say an instructor may permit limited use of generative AI for certain aspects of preparing an assignment, as long as you do not present the output as your own work.
• UCC says it varies by module. Its GenAI toolkit says what counts as appropriate use can differ across disciplines or from module to module, because it depends on what the assessment is measuring.

The practical consequence is that the answer lives in three places, and you need all three: the university policy, the programme handbook, and the assignment brief. If you are still working out how Irish assessment is built and where the marks come from, how Irish university exams actually work covers the structure this sits inside.

The ten minutes that settle it

Do this once per module in week one, and you will not have to guess in week ten.

1. Open the module descriptor and search it for "AI", "artificial" and "generative".
2. Open the assignment brief and do the same. The brief overrides your assumptions, not the other way round.
3. If neither says anything, email the module coordinator and ask, in writing, what is permitted for this assessment.
4. Save the reply. A written permission is worth more in an investigation than your memory of a corridor conversation.

What does the national guidance say?

Behind the university policies sits the National Academic Integrity Network, a sector body facilitated by Quality and Qualifications Ireland. NAIN published its Generative Artificial Intelligence guidelines in 2023, written for educators but built to be shared with students, and it says itself that the area moves fast enough that the guidelines will need regular review.

The sentence that matters to you is short. NAIN states that academic integrity is breached if students submit the products of GenAI as their own work without acknowledgement or without authorisation to use GenAI in fulfilling the task. Read the two conditions carefully. Acknowledgement alone is not enough if the use was not authorised, and authorisation alone is not enough if you never declared it.

That wording has travelled. UCC's policy uses almost the same sentence at clause 5.4, and Trinity, UCD, Galway and DCU all land in the same place by their own routes. When five institutions converge on one rule, it is usually safe to treat it as the national position.

What do Trinity, UCD and the others publish?

Two of them also tell you which tool they expect you to use. Trinity says all staff and students can reach Microsoft Copilot Chat through their institutional Microsoft 365 account. UCD says it holds an enterprise licence with Google and that staff and students can use the Gemini web app with their UCD credentials through UCD Connect, and adds a fairness rule in the same breath: staff may not require a student to use an AI tool that is not provided by UCD. Having access is not permission. It only means the data side has been sorted out for you.

When do you have to declare it?

Whenever the module allows AI and you used it. The declaration is not a confession, it is part of the submission, and leaving it out is the thing that turns permitted use into misconduct.

Trinity is the most specific about the form. Its statement says use should be acknowledged and appropriately cited, and that a citation should typically include the date of generation, the tool used and the prompts used to create the output, with verbatim text in quotation marks. It also warns that a long verbatim quotation, even a properly cited one, may be treated as poor practice in a student submission.

UCC's toolkit describes the same thing as a statement of how the tool was used, which can include the prompting process, screenshots of the output or a link to the conversation, together with a line confirming that you have reviewed the AI-generated content, that it reflects what you meant, and that you take responsibility for it. UCC publishes declaration templates for staff and students to use. University of Galway tells instructors to consider requiring a statement describing the precise nature, extent and purpose of the student's use.

DCU comes at it from a different angle. Every piece of DCU coursework carries a declaration of academic integrity, either on paper or through the electronic version on Loop, and the policy says coursework submitted without it, or where the electronic equivalent has been circumvented, will not be accepted.

What a declaration actually contains

Keep it to four lines and you will meet most of the requirements at once:

• The tool and the date. Name it and date it, as Trinity asks.
• What you asked it. The prompts, or an appendix holding them.
• What you did with the answer. Brainstormed, rewrote, checked grammar, generated practice questions.
• What is yours. A sentence saying you checked the output and stand over the content.

What is allowed, and what is not?

The table below is the pattern across the five policies, not a rule from any single one. Your module descriptor beats it every time.

One more thing the middle band does not cover: getting the tool to write the thing and then rewriting it enough to feel like yours. NAIN's guidance says a report or essay constructed through clever prompting, which anyone not taking the module could have produced, is not a basis for awarding marks or credit. That is the test a marker applies, not a word count.

What happens if you are caught?

You get a process, and it is slower and more paper-heavy than most students expect. UCC's policy is the most fully published, so use it as the worked example.

• It starts in writing. The Head of School or their nominee informs you by email to your UCC address, with detailed reasons and a copy of the work the allegation rests on.
• You answer in writing. You get a specified window to give a Personal Statement, and you may be invited to a meeting where you can bring another person for support.
• The standard is the balance of probabilities. If you do not reply or do not attend, the decision is made on the available evidence without you.
• There are three levels. Poor Academic Practice, Minor Academic Misconduct and Major Academic Misconduct. A Level 2 sanction will not exceed a mark of zero for the whole piece of work. Level 3 goes to the Records and Examinations Officer.
• It follows you within your programme. Records go on the Academic Misconduct Register, and a previous record escalates the next case by one level.
• Appeals are time limited. Ten business days from the determination, on stated grounds.

University of Galway runs a different first step that is worth knowing about. If you are suspected of intentional misconduct and have no previous breach, you are invited to a Courageous Conversation with the Academic Integrity Advisor in your discipline or school, before any formal investigation. Galway says that if you share the full details at that stage, the more serious outcomes for that category will not apply and no formal investigation takes place. You may bring another student or a Students' Union representative as a support person, though not a staff member, and the support person cannot speak for you.

If you are already in the process

Galway's FAQs make two points that students learn too late. Intent does not save you, because you are responsible for knowing the policy either way. And mitigating circumstances cannot be considered during a misconduct investigation, so if stress or a deadline pile-up is the reason, the time to raise it was before you submitted, not after. Exam anxiety and deadline pressure are worth dealing with as their own problem for exactly this reason.

Why do the references it gives you not exist?

Because the tool is not looking them up. Trinity's statement is blunt about the mechanism: GenAI tools do not store facts and knowledge, they generate outputs based on probabilities, so they make factual errors, called hallucinations, which are nonetheless convincingly presented. UCC's toolkit says the same thing in one line: GenAI is known to hallucinate and is not credible as a source.

A reference list is the worst possible use of a probability machine, because a plausible-looking citation is exactly what it is good at producing. NAIN's guidance lists compiling a list of references among the tasks these tools can already do, and points out that a traditional library or search-engine strategy finds the actual sources and lets you judge their quality. A fabricated citation in your bibliography is not a grey area. It is a claim you made about a source you never read.

Check the reference, not the sentence

If a tool has given you a citation, find it in your library catalogue or on the publisher's site before it goes anywhere near your draft. Open it. Read the page the claim is supposed to be on. If the article exists but does not say what you were told it says, that is the same problem wearing a better disguise, and it is the version that survives into a submitted essay. Trinity's statement tells students and staff alike to fact-check GenAI output and to seek out primary sources as part of ordinary academic practice.

What must never go into a chatbot?

Other people's work and other people's data. This is the part of the rules that has nothing to do with marks, and the part most students have never read.

Trinity's statement is direct: you are not allowed to use personal, private or sensitive information about colleagues or students as part of prompts or other interactions with GenAI tools, and student work, meaning submitted assessments and contributions, is not allowed to be uploaded into a third-party GenAI tool for any reason. Confidential College material is covered by the same prohibition.

UCD reaches the same conclusion through data protection law. Its guidance says personal data entered into AI tools the university has not supported or approved is a breach of GDPR and is therefore prohibited, and then makes the point that catches people out: students' academic work, their assignments and exam scripts, is the students' personal data. NAIN's guidelines told educators the same thing in 2023, warning against putting personal, private, sensitive or copyrighted material, including student work, into these tools without legal guarantees.

In practice that rules out a short list of things people do without thinking: pasting a classmate's draft in to compare it with yours, uploading a placement or clinical record to get a summary, feeding in interview audio from a research project, or dropping in a scanned handout that someone else holds the copyright in. Your rights over your own data, and what "stored in the EU" actually means is a separate question, and one worth understanding before you sign up to anything.

Detection is a question of its own, and a messier one than most students expect. What Irish universities do with Turnitin and what a similarity score means covers where that stands, including the universities that have decided not to use AI writing detection at all.

Where Notibo fits

The honest position is that a tool which turns a lecture into notes and flashcards sits in the first band of the table above, and only there. It helps you study the material. It does not produce the thing you hand in, and if you ever use it in a way that reaches the page, your module's declaration rule applies to it exactly as it applies to ChatGPT.

Notibo records a lecture, or takes an audio file you upload, and returns a transcript, structured notes and flashcards with spaced repetition. It also turns PDF, PPTX and DOCX files into notes and flashcards. It transcribes 89 languages, with the transcript in the language spoken. Pro is free for the first 14 days, with 240 minutes of recording and no card. After the trial the free plan covers 30 minutes of recording a month, and Pro is 9.99 euro a month. Uploads are capped at 50 MB per file. Ask the lecturer before you record, every time, and never upload another student's work or anyone else's personal data, because the university rules above bind you whatever tool you are using. Files are stored in the EU on Supabase in Frankfurt, while transcription and the AI notes run on US processors under standard contractual clauses. If you want the wider argument about where these tools help and where they do not, AI for studying makes it.