Yes, on conditions you meet before and after the work. The Guidelines for AI use in Graduate Research from the School of Graduate Studies and Postdoctoral Affairs (SGSPA) permit generative AI across the stages of the research lifecycle, editing included, once your supervisor(s) and/or supervisory committee approve the planned uses in advance and you have documented that approval. In the thesis you then give a citation, a descriptive statement, and where appropriate the AI outputs themselves. The PDF carries no date; we read it on November 9, 2026.
If your AI statement clearly doesn't cover what you describe elsewhere in your thesis, Esmond points it out.
SGSPA prepared these guidelines with graduate programs and university partners, under the Senate Policy on Academic Integrity. They apply to theses, dissertations, major research projects, research creation projects, comprehensive examinations, and graduate coursework. The central statement is that generative AI may be used in graduate research, from searching and outlining to drafting and editing, without counting as a departure from academic integrity, provided three conditions hold: authorized use, transparent use, and originality.
For a thesis, the first condition reads: “Students who plan to use generative AI tools in their research or writing of their thesis must, in advance, document approval for the planned uses of AI tools from their supervisor(s) and/or supervisory committee.” The order matters at Queen's: approval comes first. Before you use Esmond, or any other AI tool, for your thesis, ask your supervisor and get the approval documented.
A hypothetical example of a commentYour descriptive statement says a GenAI tool only suggested search terms, yet appendix B contains a table of themes the tool produced from your field notes. Does the statement cover that table?
A first comment on one part of your own draft, right away and without an account.
Get a free previewThe approval looks forward. It covers planned uses, so a use you had not foreseen when you asked is a new question for your supervisor, even halfway through the project.
| Condition | What SGSPA asks | What to keep |
|---|---|---|
| Authorized use | You, your supervisor(s), and where appropriate committee members agree in advance on any AI tool in the design, production, and representation of the research | The dated record of that agreement, naming each planned use |
| Transparent use | Explicit citation, a descriptive statement on AI use, and AI outputs where appropriate | Notes on each use and the outputs you relied on |
| Originality | The research reflects your own original contribution at the level of your degree | An account of the decisions you made, in your own words |
The transparency condition is the most concrete part of the document: “graduate students must provide explicit citation, a descriptive statement on AI use, and where appropriate artifacts of AI-outputs used in research and/or writing.” SGSPA ties this to the norms of your discipline, so the form can differ between fields.
Under its considerations, the guidelines repeat that AI tools used in research need proper citation and point to the APA Style Blog as one source of citation guidance. The unusual word is artifacts. We read it as the outputs themselves, such as a table a tool generated or a summary you worked from, attached where your committee finds that appropriate. Ask early which outputs they expect, so you keep them from the start.
The third condition sets the bar by degree level. For a master's degree, SGSPA expects originality in the application of knowledge. For a doctorate, it expects the ability to produce original research. The same AI use can therefore raise different questions in a master's project and in a doctoral dissertation. Ask your committee how they apply the condition to your degree before you settle on a use.
Among its considerations, SGSPA singles out work with human subjects and their data. For any AI use in which data will be stored, analyzed, and input, you must declare the use and receive clearance from the General Research Ethics Board. The guidelines also ask you to read each tool's terms of use, since those address ownership of intellectual property, and they make the user responsible for knowing the safety risks of a tool.
A central rule covers university data. Queen's guiding principles for generative AI, written for students, staff, and faculty, say: “If you are using internal or confidential data, use only Queen’s-approved AI tools like LibreChat and Microsoft Copilot with Queen’s single sign-on.” The IT Services Data Classification Standard lists “draft or unpublished documents, research, projects, and data” among its examples of Internal data. Whether your own thesis draft counts as Internal university data is not settled by these texts. Ask your supervisor before you upload it to Esmond or any other tool Queen's has not approved.
The document is plain about the risk: “Unauthorized use of generative AI tools in graduate student research at Queen’s University may be considered a breach of academic integrity and subject to investigation and sanctioning.”
The Senate Policy grounds listed at the top of the PDF explain the reach of that sentence. One ground, unauthorized content generation, covers producing academic work in whole or in part with assistance that was unapproved or undeclared, human or technological. A use your supervisor agreed to but that you never described can still cause trouble, because the policy names undeclared assistance next to unapproved assistance.
Faculties and graduate programs may add their own requirements, and graduate units may issue guidance on uses specific to their discipline. Read your unit's guidance alongside the SGSPA document.
This example is made up. A doctoral student in geography gets her committee's approval, recorded in the minutes of a September meeting, to use a GenAI tool for first summaries of 40 policy documents, which she then checks and revises. Her thesis cites the tool. Her descriptive statement gives no detail on which summaries she kept, and she deleted the tool's outputs once she had revised them. If the committee asks for artifacts, she has nothing to show. A folder of original outputs, kept from the first summary on, would have covered it.
Esmond evaluates every assignment before you submit. It never writes it.
Esmond is an AI tool that doesn't write for you. If your course doesn't allow any AI tools, don't use Esmond for that work. If you're unsure, ask your instructor or advisor.
Esmond reads a thesis draft and evaluates its content, pointing out factual errors, inconsistencies, and blind spots against the standards of your field. See how Esmond works and the sample report. At Queen's, a review of your draft by an AI tool is a use of generative AI in the writing of your thesis, so the authorized-use condition applies: ask your supervisor first and get the approval documented before you upload a chapter. Esmond is not a Queen's-approved tool, so settle in the same conversation whether your draft counts as Internal data. If you go ahead, cite Esmond and describe in your statement what it pointed out and what you did with it. If your work involves human participants, check what the General Research Ethics Board requires before any of their data goes into a tool, and read the Privacy Notice. Queen's University has not reviewed or approved Esmond.
No. SGSPA permits generative AI across the research lifecycle, editing included, if the planned uses are approved in advance and you cite and describe them. Your graduate unit may add guidance for your discipline.
Yes. The guidelines say you must document approval for the planned uses in advance. An email from your supervisor that names the tool and the task gives you that record.
For human subject research, SGSPA says you must declare and receive clearance from the General Research Ethics Board for any AI use in which data will be stored, analyzed, and input.
The three conditions apply to both. The originality condition is measured differently: application of knowledge for a master's degree, original research for a doctorate.
Read on November 9, 2026.
Get your paper evaluated before you submit