Guide · AI and academic integrity
Can I Use AI to Write My Dissertation? What Universities Allow in 2026
You can usually use AI to help with your dissertation, but not to replace your own work, and only within your university's rules. Most universities in 2026 regulate AI rather than ban it: using it for permitted tasks such as finding sources, outlining, checking citations or proofreading is often allowed, while submitting undisclosed or unauthorised AI-generated text in assessed work is treated as misconduct. Read your department's policy, disclose AI use as required, keep the argument and final wording your own, and keep a record of your process.
It is one of the most searched questions in higher education: can I use AI to write my dissertation? The honest answer is "it depends on your institution, and on what you mean by write". This guide explains what universities typically allow, where the line usually falls, how to disclose AI use, and how to protect yourself if an AI detector gets it wrong.
This is general guidance, not legal or institutional advice. Your university's and department's written policy always takes precedence.
Universities regulate AI, they rarely ban it
In 2023 many institutions reacted to ChatGPT with blanket bans. By 2026 most have moved to policies that permit some uses and prohibit others. The details vary by university, department and even individual assignment, but common patterns are clear:
- Disclosure is central. Using AI is often acceptable if you disclose it; hiding it is the problem.
- Assessed work is stricter. Rules for a dissertation, which is assessed, are usually tighter than for drafts or study notes.
- Unauthorised AI text is treated like contract cheating. Imperial College London's guidance, for example, says using AI to create assessed work may be treated as an offence similar to contract cheating unless explicitly authorised. Cambridge treats submitting AI-generated text in summative assessments without permission as misconduct.
- Departments differ. Many universities direct students to their department's own rules rather than a single university-wide policy.
What is usually allowed, and what usually isn't
| Use of AI | Typically… | Why |
|---|---|---|
| Brainstorming topics and research questions | Often allowed | Doesn't produce assessed text |
| Finding and summarising sources | Often allowed, with care | You must still read key sources |
| Organising notes and outlining chapters | Often allowed | The structure is still your decision |
| Checking citations and references | Often allowed | Improves accuracy |
| Grammar, spelling and clarity suggestions | Often allowed | Similar to editing software |
| Translating your own writing | Varies; disclose | Can change your meaning or voice |
| Drafting paragraphs you then substantially rewrite | Varies widely; disclose | Grey area at many institutions |
| Submitting AI-written text as your own | Usually prohibited | Misrepresents authorship |
| Fabricating data or references | Always prohibited | Research misconduct |
The further a task moves from helping you think and check towards producing the assessed text for you, the more likely it is to be restricted.
A practical decision checklist
Before using AI for any part of your dissertation, ask:
- What does my department's policy say? Look for a dissertation-specific or programme-specific statement, not only the university's general page.
- What does my supervisor expect? Ask directly, in writing, and keep the reply.
- Is the output going into the assessed text? If yes, the rules are stricter and disclosure matters more.
- Will the work still show my own understanding? Examiners assess your thinking, analysis and argument.
- Can I explain and defend every sentence and source in a viva or meeting?
- Have I recorded what the AI did? A process record is your best evidence.
If you are unsure, ask before you use it, not after.
How to disclose AI use
Most policies that permit AI ask you to disclose it. Princeton's guidance, for example, says AI is not a source, so its use should be disclosed rather than cited like a paper. A disclosure statement usually covers:
- Which tool you used.
- What you used it for: searching, outlining, citation checking, language editing and so on.
- How you checked its output: for example, "all references were verified against the original sources".
- That the analysis and conclusions are your own.
Example: "I used an AI research assistant to search for literature, suggest a chapter outline and check that cited sentences matched their sources. I verified every reference, rewrote all text in my own words and am responsible for the analysis and conclusions." Adapt it to your institution's required format.
Protect yourself from false AI-detection flags
AI detectors are unreliable. Research shows they produce false positives and flag non-native English writers disproportionately, and in one widely shared 2026 case a student's thesis was reportedly flagged as 98% likely AI-generated. Many institutions treat detector scores as supporting evidence rather than proof, but a flag can still start an investigation.
The best protection is evidence of your process:
- Keep drafts and version history, ideally with dates.
- Keep your notes and reading records.
- Keep a log of AI assistance: what you asked, what it produced and what you changed.
- Be able to discuss your sources and argument in detail.
Some tools help with this. CiteLoom, for example, keeps a version history of every change and can export an integrity report listing your sources, how each was verified, and every AI-assisted fix or rewrite, which you can share with a supervisor.
If your university has no clear AI policy yet
Some institutions, departments or programmes have not published specific guidance. That doesn't mean anything goes. In the absence of a policy, apply your university's existing academic-integrity rules: work submitted for assessment must be your own, sources must be acknowledged, and you must not misrepresent how work was produced. Ask your supervisor and programme leader directly, record their answer, take the most cautious reasonable interpretation, and disclose any AI use you make. If a policy is introduced later, you'll be able to show you acted in good faith.
Using AI well: a responsible workflow
- Write your research question yourself.
- Use AI to find and organise literature, then choose your sources deliberately.
- Decide the structure and argument yourself; let AI help you outline.
- If your institution permits drafting help, treat any AI text as a rough draft to rewrite, never as final text.
- Verify every citation against its source.
- Edit in your own voice and add your critical analysis.
- Disclose AI use as required and keep your records.
Three common scenarios
Scenario 1: "I used AI to find sources and check my references." At most institutions this is low-risk research support, comparable to using a database or reference manager. Read the sources you cite, verify every reference and disclose the tool if your policy asks.
Scenario 2: "AI drafted my literature review and I edited it." This is the grey area. Some departments permit AI drafting with disclosure and substantial rewriting; others prohibit any AI-generated text in assessed work. Ask your supervisor before you do this, get the answer in writing, rewrite thoroughly in your own words and disclose clearly.
Scenario 3: "I submitted AI-written chapters without saying so." This is what most policies treat as misconduct. If you are in this position, talk to your supervisor or your university's student support before submission. Coming forward early is almost always better than being discovered later.
Questions to ask your supervisor
Use these to get clear, written guidance:
- Which uses of AI does our department allow for the dissertation?
- Is AI-assisted drafting allowed if I rewrite and disclose it?
- How should I disclose AI use, and where in the dissertation?
- Are there tools the department recommends or prohibits?
- Will the dissertation be screened with an AI detector, and how are flags handled?
Keep the replies with your records. If the policy changes during your degree, you'll be able to show what you were told.
Write with citations you can defend
CiteLoom finds real papers, lets you approve the plan, drafts your review, and checks every cited sentence against the source, showing the exact quote. Pay per draft by card or M-Pesa.
Start with a free draft →Frequently asked questions
Is it illegal to use AI to write a dissertation?
It is generally not illegal, but it can breach your university's academic-integrity rules. Submitting undisclosed AI-written work as your own is commonly treated as misconduct, with penalties up to failing the dissertation.
Can my university detect AI in my dissertation?
Universities may use AI detectors, but these are unreliable and produce false positives. Examiners also notice writing that doesn't match your other work or that you can't explain in a viva.
Is using AI for my literature review allowed?
Using AI to find, organise and check sources is often allowed. Submitting an AI-written review as your own usually isn't. Check your policy and disclose your use.
Do I need to cite ChatGPT or other AI tools?
Many institutions ask you to disclose AI use rather than cite it like a source. Follow your university's and your citation style's specific guidance.
What if my supervisor allows AI but my university's policy is unclear?
Get your supervisor's permission in writing, follow any departmental rules, disclose your use and keep records of your process.