Binghamton does not ban AI in a thesis, but it makes every student say in writing what they did. The Graduate School Manual requires a declaration on generative AI in the front matter of each thesis, dissertation and culminating project, with one version for no use and one for use. University Policy 312, last revised January 23, 2026, adds that GenAI use in academic work is disclosed and that Protected Information goes into a GenAI tool only with approval. Your advisor, committee and department can expect more than that, so ask them before you start.
If your AI statement clearly doesn't cover what you describe elsewhere in your thesis, Esmond points it out.
The Graduate School Manual for theses and dissertations has a section titled Required Declaration Statements on Generative AI Use. Its instruction is short: “This statement must be placed in the front matter of the dissertation/thesis/culminating project following the formatting guide.” The manual page carries no date; we read it on November 9, 2026.
The manual does not list allowed or forbidden uses, and it does not ask for approval before you use a tool. What it requires is a record. Decisions about which uses fit your project stay with your advisor, your committee and your department, and Policy 312 sets university-wide principles on top.
A hypothetical example of a commentYour discussion chapter says the effect held in all three field sites, while Table 5 shows no significant difference at the third site. Which reading do you defend, and does the conclusion need the same qualification?
A first comment on one part of your own draft, right away and without an account.
Get a free previewThe first template is for students who used no generative AI at all: “I declare that no generative artificial intelligence (AI) tools or services were used in the research, writing, analysis, or any other aspect of this [dissertation/thesis/culminating project].”
Read the scope of that sentence closely. It covers the research and any other aspect of the work. If a generative AI tool transcribed your interviews, suggested search terms, commented on a chapter or tidied your references, the no-use version no longer fits, even if no sentence in the thesis came from a tool. In that case you use the second template, which starts with a declaration that AI tools were used and then asks for details.
The second template has four fields and a pointer to an appendix. The middle column follows the manual; the right column is our suggestion for what to keep while you work.
| Field | What the template asks | What to note as you go (our suggestion) |
|---|---|---|
| AI tools used | The names of the specific tools | The tool and, if it changes, the version or date |
| Sections or pages | Which sections or page numbers involved AI | The chapter and draft each use touched |
| Purpose | A brief account of how AI was used | One line per use, such as feedback on a chapter or help with an outline |
| Verification process | How you checked AI-generated content | What you compared the output against, and what you changed or rejected |
| Appendix | Detailed documentation of use in an appendix | Prompts, dates and outputs you may need to cite there |
The last field is the easiest one to leave thin: “Verification Process: [Briefly explain how AI-generated content was verified]” It assumes you did check, and it asks you to say how.
For a reference list, that means opening each source a tool suggested. For a summary of literature, it means comparing it with the articles. For feedback on a draft, it means deciding which comments were right and which you set aside, with your own reasons. A statement that names a tool and a purpose but no check leaves the most obvious committee question unanswered.
Policy 312 is Binghamton's Generative AI Policy, filed as an Information Technology and Information Security policy, with the Center for Learning and Teaching as responsible office. It was first approved in April 2025. One of its principles applies directly to a thesis: “Be Transparent: Clearly disclose when GenAI tools are used in academic work, research, or other university activities.”
The policy also asks users to keep to Binghamton's academic integrity policies and New York State IT policies, to check GenAI output for accuracy and bias, and to treat that output as advisory, with people reviewing any decision based on it. It supports the use of GenAI tools by students, faculty and staff, within those principles.
The data rule in Policy 312 is the strictest sentence in the Binghamton texts: “The input of personal, sensitive, or confidential information (Protected Information) into any Generative AI (GenAI) tool is prohibited unless approval has been granted.” Approval comes from Information Security, after a documented risk review.
The policy does not name thesis chapters. It applies to the information in them. Interview transcripts with names, health details of participants, student records or data under a confidentiality agreement fit the policy's words personal, sensitive or confidential, and so does a chapter that quotes them. Remove or replace those details before a chapter goes into any GenAI tool, or get the approval first. Whether your own unpublished draft counts as confidential information is not settled by these texts, so ask your advisor before you upload it to Esmond or any tool Information Security has not approved.
This example is made up. A master's student in anthropology records twelve interviews and runs them through an AI transcription service. She considers that a technical step and signs the no-use declaration, because every sentence of the thesis is her own. Two problems follow. The declaration covers the research and any other aspect, so the transcription was a use she should have declared, with a verification note on how she checked the transcripts. And the recordings contained participants' names and workplaces, which made them Protected Information under Policy 312, so the upload needed approval from Information Security before it happened. Her advisor would have caught both in one conversation.
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; how Esmond works and the sample report show the kind of comments it returns. At Binghamton that counts as generative AI use, so you would use the AI-use version of the declaration: Esmond as the tool, the chapters you sent, feedback on content as the purpose, and how you weighed each comment as the verification. Check first with your advisor and committee that they accept this use, follow any department rule, and leave Protected Information out of the file. Whether the unpublished draft itself counts as confidential is not settled by the Binghamton texts, so ask your advisor about that before you upload it. See whether you can use Esmond for the general question. Binghamton University has not reviewed or approved Esmond.
No. The Graduate School Manual asks for a declaration of AI use, and Policy 312 supports responsible use with disclosure. Your advisor, committee or department can still set limits.
No. The manual calls it a required statement in the front matter, with one version for no use and one for use.
The no-use version covers the research and any other aspect of the work, so a transcription tool is a use to declare.
Not if it contains personal, sensitive or confidential information, unless Information Security has approved that tool for it. Policy 312 says so.
Read on November 9, 2026.
Get your paper evaluated before you submit