AI Policies for Research Papers, Theses, and Dissertations Explained

Graduate researchers using AI face a much more complex set of rules than undergraduates.

An undergraduate typically follows one policy, the course syllabus. But a graduate student working on a thesis or dissertation often has to navigate three different layers of expectations at the same time:

• University or graduate school policies
• Journal or publisher requirements
• Funding agency guidelines

And these rules don’t always align.

Getting it wrong isn’t just a technical mistake, it can affect publication, funding, or even academic standing.

US University AI Policy Repository → Browse AI policies by institutions.

Key Takeaways

• AI rules at the graduate level are becoming formalized and increasingly strict
• Journals universally require disclosure and prohibit AI authorship
• Using AI tools with sensitive or unpublished data can create serious compliance risks
• Researchers must follow multiple policies at once not just their university’s

University Policies: Higher Stakes, Stricter Expectations

At the graduate level, expectations are higher because the work is expected to be original and defensible.

A dissertation isn’t just another assignment; it’s a contribution to knowledge. That’s why universities are stricter about how AI can be used.

Across institutions, a few patterns are emerging:

• Using AI to generate text, arguments, or analysis without disclosure is treated as misconduct
• AI tools are considered assistive, not authoritative
• Transparency is expected whenever AI contributes to the work

In many programs, students are now expected to include a clear AI disclosure statement, typically in the methods section, acknowledgements, or preface. This usually includes:

• The tool used
• How it was used
• The extent of its contribution

In some cases, students may also be asked to retain prompt logs or interaction records.

For exams, defenses, or formal assessments, the default assumption is still restrictive:
AI is not allowed unless explicitly permitted.

The Data Privacy Risk Most Researchers Overlook

One of the biggest risks with AI in research isn’t writing, it’s data handling.

Many researchers unknowingly expose sensitive information when using AI tools.

Uploading the following into public AI systems can violate institutional or legal rules:

• Unpublished research findings
• Participant data
• Confidential interviews
• Proprietary datasets

Once data is entered into a public tool, you lose control over how it may be stored or used.

This can lead to:

• Ethics violations
• Data protection breaches
• Intellectual property risks

The safest approach is simple:

• Only use AI with public or fully anonymized data
• Use institution-approved tools when working with sensitive material

Journal and Publisher Policies: What Happens at Publication Stage

When your research moves toward publication, a new set of rules applies.

Across major academic publishers, there is strong consistency on a few points:

AI cannot be an author

This is universal. AI tools cannot be credited as authors because they cannot take responsibility for the work.

AI use must be disclosed

If AI has contributed beyond basic editing, disclosure is required.

However, policies differ in how they define that boundary:

• Editing assistance (grammar, clarity, formatting) is often allowed without formal disclosure
• Content generation (ideas, summaries, writing) must be disclosed

Some journals require:

• A formal declaration section
• Method-level explanation of AI use
• In some cases, citation-style acknowledgment

The key distinction: assistive vs generative use

This is becoming the central idea in most publisher policies:

• Assistive AI → improves your work
• Generative AI → creates content

If AI is generating ideas, structure, or text even if you edit it heavily, it usually falls into the second category.

And that requires disclosure.

Funding Agencies: The Third Layer Most People Forget

If your research is funded, there’s another layer of rules that applies.

Funding agencies are increasingly setting their own AI guidelines, especially for:

• Grant proposals
• Research reporting
• Peer review processes

For example:

• Some agencies restrict AI use in proposal writing
• Others require transparency in how AI is used
• Certain review processes prohibit AI tools entirely

These rules apply independently of your university or journal.

That means:

• Following your university’s policy is not enough
• You must check the latest guidance from the funding body before each submission

A Practical Way to Stay Compliant

Managing three sets of rules can feel overwhelming, but it becomes easier if you treat each stage separately.

For your thesis or dissertation:

• Follow your university or program guidelines
• Disclose all meaningful AI use
• Keep records of how AI was used

For journal submissions:

• Check the specific journal’s policy before submitting
• Prepare a disclosure statement, even if unsure
• Err on the side of transparency

For grant applications:

• Review the funding agency’s current rules
• Don’t assume past guidance still applies
• Be cautious with AI in proposal writing

A Simple Rule That Works Across All Three

When in doubt, disclose.

Over-disclosure rarely causes problems.
Under-disclosure almost always does.

Conclusion

AI is now part of the research process, but it comes with layered responsibilities.

Graduate researchers aren’t just following one policy. They’re navigating expectations from institutions, publishers, and funding bodies simultaneously.

While the details vary, the underlying principle is consistent:

AI can support your research, but it cannot replace your intellectual responsibility for it.

👉 US University AI Policy Repository → Explore the AI policies