Can You Use ChatGPT for Your Thesis? What the Rules Say in 2026

Short answer: usually yes for some tasks, rarely for all of them, and almost always with disclosure. The rules come from three places, your university, your programme or supervisor, and the research integrity bodies in your country. This guide explains what those rules tend to say and how to stay on the right side of them.

What is usually allowed

• Language help. Grammar, readability and translation of your own text. Many policies treat this like a spell checker, though some still want it declared.
• Understanding. Asking a tool to explain a method, a statistical test or a difficult paper.
• Brainstorming and structure. Outlining chapters, generating search terms, suggesting counter-arguments.
• Working with your own sources. Summarising and questioning papers you have read, as long as you check the answers against the paper.

What is usually not allowed

• Generating the thesis text and presenting it as your own, especially analysis, argument and conclusions.
• Fabricating data or sources. AI tools invent references. A fake citation in a thesis is misconduct whether or not a tool wrote it.
• Uploading restricted material. Interview transcripts, patient data or unpublished results to a tool that may train on them, when your ethics approval or data agreement forbids it.
• Hiding the use. Most policies allow more than students expect, but almost none allow undisclosed use.

Examples of published rules

Policies differ, so read yours. These three show the range.

• The University of Helsinki allows AI as a learning support and asks students to disclose use, or non-use, in written work of one page or more, with the scope, purpose, model, verification and effect on learning.
• Finland’s research integrity body TENK says the researcher stays responsible for everything in the work, that substantial AI use should be described, and that purely technical help generally need not be.
• The European Commission’s living guidelines ask researchers to disclose substantial generative AI use, to remain accountable for the output, and to be careful with sensitive data.

How to find your own rules

1. Search your university’s website for “generative AI” and “academic integrity”. Most have a page, and many updated it in 2025 or 2026.
2. Check your programme or faculty handbook; faculties often add stricter rules than the university.
3. Ask your supervisor in writing what they consider acceptable for your thesis, and keep the answer.

How to disclose

A disclosure is usually a short paragraph in the preface, methods or an appendix saying which tool you used, for what, on which parts, and how you checked the output. Our free AI-use disclosure statement generator writes one in the format your university or publisher asks for. If you quote AI output directly, cite it as well; see how to cite ChatGPT.

Use it in a way you can defend

Keep a log of what you used and when. Keep the AI’s work separate from yours until you have checked it. Verify every reference against the real paper. And when in doubt, disclose more rather than less. Examiners rarely penalise honest, well-described use; they do penalise surprises.