University AI Rules: What You Can Safely Submit

Using generative artificial intelligence for revision is often allowed, but assessed work may have different limits. University-wide policy, course instructions, and tutor discretion can all affect what you’re allowed to submit.

The safest approach is to treat every assessment as its own set of instructions. Your course handbook, assessment brief, marking criteria, and tutor’s announcements form the relevant institutional guidance. They matter more than broad claims that “AI is allowed at university”.

Interpret that policy framework alongside the specific assessment instructions before opening ChatGPT, Grammarly, Copilot, or another tool.

Key Takeaways

Your own module guidance decides whether AI is permitted for an assessment.
Generative AI tools can often support private study, such as explaining a concept or creating practice questions.
Do not submit AI-generated writing, code, images, data, or analysis unless the brief clearly permits it.
Acknowledging AI use does not make unauthorised use acceptable.
Never paste confidential data, unpublished research, personal data, or university materials into a public AI tool.
Check every fact, quotation, reference, calculation, and source against the relevant citation requirements.
Keep a record of prompts and outputs if your tutor permits AI-assisted work.

A disclosure statement is not a permission slip. If the brief bans generative AI, declaring that you used it can still lead to an academic integrity case.

Why University AI Rules Differ Between Assessments

Universities have a duty to assess your own knowledge, judgement, and skills. However, the skill being assessed changes, and permitted use may depend on the assessment’s learning outcomes. An exercise using generative artificial intelligence may fit a software engineering module. The same approach could defeat the purpose of a reflective essay or a closed-book examination.

That is why a university-wide policy often provides broad principles within a policy framework. Departments and tutors may then set practical limits through instructional guidelines, reflecting faculty expectations for each assessment. The Quality Assurance Agency has called for clear and consistent institutional policies, but UK universities still apply them differently across assessments.

The Assessment Brief Is Your First Authority

Start with the assessment brief, then check the course syllabus, module handbook, and virtual learning environment. Look for sections called “assessment guidance”, “use of generative AI”, “permitted tools”, or “assessment conditions”.

The University of Edinburgh tells students to check school guidance and individual assessment criteria because generative AI may be prohibited for certain modules or tasks. Its guidance on AI and academic integrity also shows why you cannot assume that one course rule applies everywhere.

If the instructions are unclear, email your tutor before doing the work. Ask a focused question: “May I use ChatGPT to produce practice questions, but not include its wording in my submission?” Saving the reply is part of your student responsibilities.

A Practical Permission Matrix

Use this quick guide while reading your brief and checking which generative AI tools are allowed.

Assessment Instruction	What You Can Usually Submit	What You Should Do
AI prohibited	Work created without generative AI help	Avoid AI tools for drafting, rewriting, coding, translation, or images
AI allowed for planning	Your own final work	Use AI only for the named planning activity and keep notes
AI allowed with disclosure	Work that follows the authorized use limits	Add the required declaration and retain prompts if requested
No guidance given	Only work you know is your own	Ask your tutor before using AI in any part of the assessment

When in doubt, choose the stricter option. A short email is far less stressful than a formal investigation.

Safe AI Uses for Private Study

Many students use AI productively before an assessment starts. Generative AI tools can support learning without producing work you submit as your own. The key distinction is simple: use them to learn, not to write assessed work for you.

The University of Birmingham permits generative AI as a study aid for personal learning and research, while its rules restrict using AI output in assessments unless the relevant code or brief authorises it. That boundary is common across university AI rules. A particular module or assessment may set stricter limits, so don’t assume a general study allowance applies to assessed work.

Use AI to Test Your Understanding

Ask for a plain-language explanation of a difficult theory, then close the chat and explain it yourself in your notes. You can also request practice questions, flashcard prompts, a comparison of concepts, or feedback on whether your revision plan covers the module topics. This helps build AI literacy, as you question, check, and restate the answer independently.

For example, a psychology student could ask for ten practice questions on classical conditioning. The student should then answer them independently and check each explanation against lecture materials.

This works best alongside active recall and spaced review. For more ideas, see these tips on using AI for effective studying.

Use It to Find Starting Points, Not Evidence

AI can suggest search terms, names of theories, or possible counterarguments. It can help you make an initial research plan when a blank document feels daunting.

However, don’t treat its output as a source. Find the original journal article, official report, book chapter, or dataset. Read it yourself, then cite that source according to your department’s referencing style.

A useful habit is to ask AI for keywords rather than answers. “What search terms would help me find peer-reviewed studies on sleep and memory?” gives you a direction without handing over your academic judgement.

What You Should Never Submit Without Permission

Submitting writing generated by generative AI tools as your own is the clearest risk. The restriction also covers code, calculations, data, references, translations, visual designs, spreadsheet formulas, lab reports, images, and audio.

Photo by RDNE Stock project

AI-Written or AI-Rewritten Assessments

Do not paste your essay into artificial intelligence tools and submit a heavily rewritten version unless the assessment explicitly allows it. The same applies when a chatbot writes your introduction, conclusion, literature review, or reflective account.

Translation assistance and image generation may also be prohibited, even when the final submission looks original.

A spellchecker that highlights a typo differs from artificial intelligence tools that change your argument, sentence structure, voice, or analysis. University of Kent’s policy draws a similar line, warning against work substantially changed by AI beyond ordinary spelling and grammar support.

Submitting generated work without permission may be treated as academic dishonesty, not merely a minor technical breach. AI-generated content can also be easy to spot for human reasons. It may use vague claims, invented references, an unfamiliar writing style, or conclusions you cannot explain in a seminar.

AI-Generated Data, Citations, and Code

Never submit fabricated survey results, made-up interview quotations, false laboratory observations, or AI-invented references. These actions can turn a weak assignment into a serious research integrity issue.

Copyright and intellectual property concerns can also arise with AI-generated images, visual designs, lecture materials, or employer-owned content. They can matter even when the final submission looks original.

Code needs the same care. If an assessment allows AI-assisted programming, student responsibilities include understanding, testing, and explaining every line. Check whether the brief, university, or placement provider has authorised them before using unapproved tools. Also check software licensing to confirm that generated code, assets, or images may legally be reused. If it doesn’t, submitting generated code is no different from submitting another person’s work.

City St George’s makes clear that AI use which breaches its academic integrity policy may be treated as misconduct. Read your university’s definition of misconduct before assuming that a small AI contribution will be ignored.

Protect Your Data and Other People’s Work

Public AI services aren’t private filing cabinets. Prompts may be stored, reviewed, or used to improve a service, depending on the tool, account settings, and its data privacy practices.

Don’t upload information you wouldn’t post on a public noticeboard. This includes your university login details, student number, identifiable participant data, placement records, employer records, unpublished dissertation findings, and tutor feedback. Treat these materials as confidential data.

Check the tool’s retention and account settings before using it. Review its data privacy terms and consider re-identification or exposure risks in a risk assessment. Follow your university’s security policies.

Treat Confidential Material With Care

Health, social work, law, education, business, and research projects often involve confidential data. Even changing a name may not make a case study anonymous if its circumstances identify a person.

University materials can also be protected by intellectual property rights or contractual terms. Lecture slides, paid course readings, unpublished data, employer documents, and generated assets shouldn’t be copied into unapproved tools. Check software licensing terms before using external tools or generated assets for assessed work.

The University of Cambridge’s AI and data protection guidance is a helpful reminder that institutional work needs proper data handling. If your university provides an approved enterprise tool, it may offer secure access and stronger controls than a public account. That doesn’t automatically permit uploading assessment material. Use it only where your course permits it, and check current institutional and provider settings.

Disclose AI Use in the Required Format

If AI is allowed, follow the requested format exactly. Authorized use of generative AI tools must stay within the assessment brief’s limits, and disclosure doesn’t make prohibited use acceptable. Some tutors ask for a short declaration. Others want an appendix containing prompts, outputs, and an explanation of how you used the tool. A few modules may provide a template.

Your statement should be factual and support transparent usage. Name the tool, say what you used it for, and give the date you accessed it. Explain how it affected your work. Don’t claim the tool supplied reliable evidence when it only gave you a rough starting point. Honest disclosure supports ethical standards, but it can’t legitimise prohibited work.

Put the Real Sources in Your Reference List

Reference the book, article, report, dataset, or website you actually relied on. Don’t cite a chatbot as though it were a scholarly authority.

For example, if AI points you towards a report by the Office for National Statistics, read the report and cite the ONS publication. If the tool can’t provide a working source, remove the claim.

Some universities ask students to acknowledge AI use but don’t want AI output included in a conventional reference list. Others use a prescribed citation style. Follow the institution’s compliance standards for its prescribed declaration, record retention, and citation requirements. These processes differ between institutions.

Keep a simple record while you work:

Save the exact prompts that influenced permitted work.
Note which outputs you rejected, checked, or rewrote.
Record the tool name, version if available, and date of access.
Keep copies until grades and any review period have passed.

Check AI Output Before It Damages Your Work

Generative AI predicts plausible language. It doesn’t reliably distinguish a well-supported fact from convincing fiction. This is why it can invent authors, articles, quotations, statistics, legal cases, and web links. These invented details are often called hallucination errors, but the label itself doesn’t show whether an output is reliable.

The Open University’s student guidance on generative AI warns about these pitfalls while recognising AI’s learning potential. Treat every output as unverified notes, then check AI study notes for wrong facts before anything reaches an assignment.

Use a Four-Part Verification Habit

First, open the original source rather than trusting a summary. Next, check that the author, publication date, title, and quoted claim all match.

Then compare key facts with a second reliable source, especially for statistics and controversial claims. Finally, ask whether you can explain the point in your own words and defend it in a tutorial.

Together, these checks build AI literacy: you verify sources, compare evidence, and defend a point in a tutorial.

For research-heavy work, this guide to checking AI-assisted research can help you separate useful leads from unreliable claims.

A Five-Minute Check Before You Submit

Pause before uploading your work. Read the assessment brief and any instructor instructions again. Check them against the university’s general policy, rather than relying on what you remember from week one.

Confirm whether the assessment allows generative AI at all, even if the wider university policy permits it.
As part of your student responsibilities, check that every submitted sentence, figure, calculation, and code section meets the stated rule.
Remove any invented or unverified citations, then replace them with original sources.
Add the required AI declaration, prompt log, or appendix in the format specified by the brief or your instructor.
Make sure you can explain every decision and claim without opening an AI chat.

This check protects more than your mark. It doesn’t guarantee acceptance, but it builds AI literacy, helping you understand, verify, and defend your work. It also protects the learning you’re paying time and tuition fees to gain.

Final Thoughts on Responsible AI Use

The safest submission is work you understand, can defend, and completed within the applicable assessment rules and current institutional policies. AI can support revision and early research, but it cannot take responsibility for your academic judgement.

Read each brief carefully, protect confidential information, verify all evidence, and disclose permitted AI use honestly and responsibly, following ethical standards. Your own thinking must remain visible in the work you submit.

Frequently Asked Questions About University AI Rules

What Happens If I Submit AI-Generated Work Without Permission?

Your university may handle unauthorised AI use under academic-misconduct procedures, its student code of conduct, or both. Possible outcomes include a warning, reduced mark, failed assessment, resubmission requirement, or formal disciplinary process. The response varies by institution, the amount of unauthorised work, and whether there have been previous concerns.

Can I Use ChatGPT to Improve Grammar?

Only if your assessment guidance allows it. Basic spelling and grammar correction may be acceptable, but a chatbot can substantially rewrite ideas and alter your voice. Ask your tutor if the brief doesn’t state what level of editing is permitted.

Can I Use AI to Find Sources for an Essay?

You can often use it to generate search terms or possible leads during private research. However, open and read every original publication yourself. Never cite a source you haven’t accessed and checked.

Do I Have to Declare AI Use If I Only Used It for Planning?

Follow your module’s instructions, as course and instructor requirements can differ. Some assessments require disclosure of any generative AI use, including brainstorming or outlining. Others only require it when AI affected submitted content. When the guidance is unclear, disclose it or ask your tutor.

Are AI Detectors Proof That I Cheated?

No. AI detectors can make mistakes, so they shouldn’t be treated as conclusive proof on their own. Keeping drafts, notes, research records, and any permitted AI prompt history helps show how you developed your work.