Penn State's graduate integrity policy, GCAC-805 (adopted by Graduate Council in February 2025, in effect since fall 2025), covers dissertations and other milestones. It prohibits tools that the people responsible for the work have not permitted, and it leaves the limits on generative AI to them. On November 9, 2026 we found no Graduate School rule that spells out AI use in a thesis, so the details sit with departments. Two engineering departments publish them: Engineering Science and Mechanics asks you to discuss any AI use with your advisor and committee, and Architectural Engineering allows editorial help. Ask your advisor first, and check whether your department wants each tool named in the document.
If your AI statement clearly doesn't cover what you describe elsewhere in your thesis, Esmond points it out.
GCAC-805 is the academic integrity policy of the J. Jeffrey and Ann Marie Fox Graduate School. It lists actions that count as violations, and it opens that list with a condition: “Unless explicitly permitted by an instructor supervising the work, the following actions are prohibited in any academic context”. The policy says this includes assessments outside courses, and it names qualifying exams, comprehensive exams and dissertations.
Using unauthorized tools and technologies is on the list. For generative AI, the policy adds one sentence: “The limits of use of generative artificial intelligence on examinations and other formative or summative assessments must be defined by the course instructor or those responsible for the assessment.” For a thesis, that means your advisor and committee. The rule works by permission: a tool nobody allowed counts as unauthorized. So ask your advisor first before you use Esmond, or any other AI tool, for your thesis or dissertation.
GCAC-805 does not describe how to disclose AI in a thesis. It asks programs to make expectations clear before a milestone, including on technology and editorial help. The two departments below are examples of that. They are not university-wide rules.
A hypothetical example of a commentYour methods chapter reports a mesh convergence study with three mesh sizes, but the results table shows only two. Was the third run left out, and does the stated error bound still follow from the data you show?
A first comment on one part of your own draft, right away and without an account.
Get a free previewThe ESM Graduate Programs Guide (fall 2026 and spring 2027 edition, revised August 27, 2026) has an ESM Generative AI Policy. Its first requirement is a conversation: “Graduate students are required to discuss any use of generative AI in their dissertation or thesis with their adviser and committee.” The guide says the qualifying exam, the comprehensive exam, the scholarly paper and the thesis are expected to be the student's original work, so a tool may support the work and may not replace it.
The disclosure rules are detailed. “As such they must indicate which generative AI tool they have used and how they have used it.” In practice the guide asks for an in-text citation right after any sentence with information copied from or created by a tool, copied text inside quotes, a note when images or graphics were copied or generated, and a list of the tools as a Sourced Tool at the end. It refers students to the Penn State Libraries guide on citing generative AI.
ESM also gives a fixed text for the acknowledgments of a thesis or scholarly paper. It names the tools, says what they were mainly used for, and states that the author takes full responsibility for the intellectual content and reviewed every AI-assisted section for accuracy. A shorter sentence covers general uses such as a language check. If you used no AI tool, the guide gives a statement for that too, so an ESM thesis always carries one of the two versions.
For the qualifying and comprehensive exams, the same statement goes on a separate page. The policy names Copilot as one of the tools it covers, along with all other generative and AI-based tools.
The Architectural Engineering Graduate Student Handbook (2026-27, dated August 2026) sets a different line. “Generative AI tools may be used for editorial assistance, such as grammar, style, and formatting; AI should never be the primary author.” The student is responsible for the first draft and for checking every result a tool gives.
It also allows AI as a research tool under two conditions: “They may also be used as a complementary research instrument where the student's advisor or the responsible instructor has approved that use, and the use is documented in the methods of the resulting work.” The research questions, hypotheses, interpretation of data and literature, and conclusions of a thesis or dissertation must be the student's own. The handbook adds that “Instructors and committees may impose requirements more restrictive than this paragraph for a specific course or assessment.”
| Question | Engineering Science and Mechanics | Architectural Engineering |
|---|---|---|
| Who must know before you use AI | Your advisor and committee, for any use | Your advisor or instructor, for research use |
| Editing and grammar | Allowed with disclosure; named in the statement | Allowed as editorial assistance |
| Where the disclosure goes | In-text citations, a Sourced Tool list and an acknowledgment statement | Research use goes in the methods |
| Stricter rules | Set by the department policy | Committees may add them |
ESM puts it directly: “The use of generative tools in a way that is not deemed acceptable by this policy, can constitute a violation of PSU Academic Integrity (policy GCAC-805) and may result in sanctions including suspension or dismissal.” Under GCAC-805, violations range from minor to major, and the Fox Graduate School's sanctioning guidelines apply. For a milestone evaluated by a committee, such as a final oral defense, the Architectural Engineering handbook describes how that formal process applies.
Penn State has many graduate programs, and this page quotes two. GCAC-805 asks every program to state its expectations before a milestone, so look for them:
This example is made up. A master's student in Architectural Engineering uses an image classifier to sort 3,000 photos of building facades by damage type. His advisor approved this research use in an email. In the thesis he mentions the tool in the acknowledgments and thanks it for saving time. The methods chapter describes the photo set and the damage categories and says nothing about how the sorting was done. The handbook asks for research use to be documented in the methods. The approval is there. The record of it sits in the wrong section, and a reader cannot check how the categories were assigned.
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 evaluates the content of a draft from within your field and points out factual errors, inconsistencies and blind spots. See how Esmond works and a sample report. Under GCAC-805, the people responsible for your thesis set the limits on generative AI, so ask your advisor first before you use Esmond for this thesis or dissertation. In ESM, discuss it with your advisor and committee as the policy requires. If they agree, name Esmond and what you used it for in the way your department asks, for instance in the ESM acknowledgment statement. See also whether you can use Esmond. Penn State has not reviewed or approved Esmond.
Not as such. GCAC-805 prohibits tools that were not permitted and leaves the limits on generative AI to those responsible for the work. Ask your advisor and committee before you use it.
Under GCAC-805, the people responsible for the assessment, which for a dissertation means your advisor and committee. Your department may add rules.
Yes. The ESM guide gives an acknowledgment statement for theses that used AI tools and a separate statement for theses that used none.
The AE handbook allows editorial assistance such as grammar, style and formatting. Committees may set stricter rules.
Read on November 9, 2026.
Get your paper evaluated before you submit