This is general educational guidance, not legal advice. Always follow your university's policies and your committee's instructions.
A vague artificial intelligence policy creates a serious practical problem for doctoral students. You may receive one instruction from your university handbook, another from your department, and a third from your dissertation chair. Each may appear reasonable. Together, they can leave you uncertain about what is permitted.
That uncertainty is not a reason to panic. It is a reason to create a careful record.
Recent reporting illustrates the problem. A September 2026 review by The Harvard Crimson found that Harvard College concentrations differed substantially in their treatment of AI in senior theses. Some prohibited AI use. Others allowed limited research or analysis support but not writing. Many left decisions to individual advisers, while other concentrations were still developing policies. The review concerned undergraduate theses, not a universal doctoral rule. However, it demonstrates how quickly expectations can diverge across academic units.
For doctoral students, the most cost-effective protection is not guessing what your institution intended. It is documenting what you understood, what you asked, what you were told, and how you completed the work.
Start with the policy hierarchy
Before using any AI tool in dissertation work, identify every policy that may apply:
- University-wide academic integrity and technology policies.
- Graduate school or doctoral handbook requirements.
- Department or program policies.
- Dissertation committee or chair instructions.
- Research ethics, privacy, and data-security requirements.
- Publisher or journal requirements if your dissertation includes manuscripts.
Do not rely on a search result, an old PDF, or a classroom policy that does not address dissertations. Locate the current version on an official university website or student portal.
Record:
- The policy title.
- The issuing office.
- The publication or revision date.
- The URL or document location.
- The date you accessed it.
- The specific section relevant to your work.
Policies can change during the academic year. If you began writing under one version and submitted under another, the timeline may matter. Keep a copy of each policy that governed your work at the time.
University guidance often places responsibility on the student to follow local academic-integrity and data-protection rules. For example, Harvard’s AI guidance directs users to consider security, privacy, academic integrity, and applicable school or unit policies. That layered structure is common. Your institution may use different terminology, but the principle remains important: a general university statement may not answer every dissertation-specific question.
Ask for written clarification before proceeding
A verbal conversation with your chair may be helpful. It is not always sufficient as your only record.
If your chair says that limited AI use is acceptable, send a concise follow-up email. Confirm your understanding and ask the chair to correct anything inaccurate. Include the exact tasks you plan to perform.
For example:
Thank you for discussing AI use in my dissertation. My understanding is that I may use [tool name] for [specific permitted task], but not for [restricted task]. I will not upload confidential, identifiable, unpublished, or participant data. I will retain a record of my use and include any disclosure required by the program or graduate school. Please let me know if this understanding is incorrect or if I should obtain approval from another office or committee.
Request clarification on specific activities:
- Brainstorming or outlining.
- Grammar and spelling support.
- Academic proofreading.
- Literature searching or summarization.
- Coding assistance.
- Data cleaning or statistical analysis.
- Translation.
- Rewriting or paraphrasing.
- Drafting dissertation sections.
- Developing tables, figures, or references.
This level of detail matters. “AI is allowed” is not a complete instruction. A tool permitted for spelling support may not be permitted to generate an interpretation of results. A tool allowed for coding assistance may not be approved for processing identifiable participant data.
If the chair’s answer conflicts with a written program policy, do not silently choose the more permissive interpretation. Ask the chair, program director, or graduate studies office to reconcile the instructions in writing.

Keep a contemporaneous AI-use log
If AI use is permitted, maintain a log while the work occurs. Do not attempt to reconstruct it months later.
Your log can be a spreadsheet or secure research note. Include:
| Date | Tool and version | Task | Dissertation location | Output used? | Verification |
|---|---|---|---|---|---|
| October 3, 2026 | Tool name/version | Grammar suggestions | Chapter 2, pages 14–18 | Selected edits only | Reviewed against original meaning |
| October 5, 2026 | Tool name/version | Code explanation | Analysis notebook | No direct output retained | Tested code independently |
Also preserve, when permitted and appropriate:
- Prompts.
- Outputs.
- Screenshots or exported conversations.
- Notes explaining what you accepted, rejected, or changed.
- Verification sources.
- The date and time of use.
- The tool’s model or version.
- Any disclosure language required by your institution.
Keep the log secure. It may contain your research ideas or sensitive project information. Do not place confidential prompts or participant data into a third-party system merely to create documentation.
When in doubt, disclose more rather than less. A conservative disclosure is not an admission that your dissertation lacks originality. It is evidence that you understand your responsibility for transparency.
Protect confidential and participant data
Never paste confidential research material into a public AI tool unless your university has specifically approved that environment for the relevant data.
This includes:
- Names and contact details.
- Interview transcripts.
- Medical or educational records.
- Identifiable survey responses.
- Unpublished datasets.
- Research-participant narratives.
- Proprietary employer or organizational information.
- Draft manuscripts subject to confidentiality agreements.
- Passwords, access credentials, or internal documents.
Removing a name may not be enough. A combination of dates, locations, occupations, quotations, or unusual characteristics can identify a participant.
Use approved institutional systems when AI assistance is authorized for sensitive work. If you are uncertain whether a tool is approved, ask your research compliance office, information security office, or principal investigator. The inconvenience of asking is minimal compared with the cost of a privacy incident, ethics concern, or breach of a data-use agreement.
Preserve evidence of your process
Your dissertation should show your intellectual development. Preserve that development.
Use version history in Google Docs, Microsoft Word, Overleaf, Git, or another appropriate system. Save dated copies of:
- Research questions.
- Literature review notes.
- Methodology drafts.
- Coding decisions.
- Statistical output.
- Memos and analytic reflections.
- Committee feedback.
- Revised chapters.
- Data dictionaries and codebooks.
- Reference-management libraries.
- Drafts submitted to your chair.
Process evidence is especially valuable when policies are unclear. It can help demonstrate that your ideas, decisions, analysis, and conclusions developed through your own scholarly work.
Do not create artificial evidence after the fact. Maintain ordinary, honest records as you work. Your goal is not to produce a performance for an automated detector. Your goal is to maintain a credible research trail.

Treat disclosure as part of academic integrity
If your institution allows AI assistance, follow its required disclosure format and location. A disclosure may belong in an appendix, acknowledgments, methods section, author note, or separate form.
A useful disclosure generally identifies:
- The tool or tools used.
- The dates or stage of use.
- The purpose of the assistance.
- The chapters or components affected.
- The extent of your revisions.
- How you verified accuracy.
- Which intellectual contributions remained your own.
Do not claim that AI “wrote nothing” if it generated text that you substantially revised. Do not describe substantive analytic assistance as mere proofreading. Precision is safer than minimization.
A disclosure does not automatically make every use permissible. Your institution may prohibit certain activities even when they are disclosed. The governing policy and committee instructions remain decisive.
Complete a pre-submission compliance review
Before submitting a proposal, chapter, article manuscript, or final dissertation, conduct a short audit:
- Did I read the current policy?
- Did I save the policy version that applied when I wrote the relevant material?
- Did I obtain written clarification for ambiguous uses?
- Did I maintain an AI-use log?
- Did I protect participant and confidential data?
- Did I preserve version histories and research notes?
- Did I verify every citation, quotation, statistic, and factual claim?
- Does my disclosure match my actual use?
- Can I explain my methods and conclusions without relying on an AI tool?
- Have I followed my chair’s and committee’s latest instructions?
A qualified dissertation coach can help you organize this process without taking ownership of your scholarship. The ethical role of a coach or editor is to guide, question, clarify, and improve presentation. It is not to write your dissertation or replace your independent judgment.
Likewise, academic editing services and academic proofreading should be used to strengthen clarity, organization, grammar, and formatting in your work, not to conceal unauthorized AI-generated content or substitute for your research contribution. If APA compliance is a concern, review the principles in this guide to common APA mistakes and how an academic editor addresses them.
The safest response to ambiguity is documentation
An unclear policy does not mean your dissertation is unprotectable. It means you must be more deliberate.
Ask precise questions. Keep the answers. Record the policy version. Protect your data. Preserve your drafts. Disclose conservatively. Maintain command of every argument, method, result, and conclusion.
The return on that effort is substantial. A few minutes spent documenting an instruction can save weeks of uncertainty later. A careful AI-use log can be more valuable than trying to remember your process before a defense. A secure workflow protects both your research participants and your academic standing.
Panther Academic Editing’s editors are experts in academic editing, which is a specialty in the editing field. The company provides guidance-focused support and a guarantee on its editing work that is unparalleled in the industry. That guarantee does not promise committee approval or override university policy. Your dissertation remains your work, your responsibility, and your scholarly contribution.
When expectations are changing, clarity is indispensable. Build a record that reflects the care, expertise, timeliness, and integrity your doctoral work requires.





0 Comments