AI & Academic Integrity

How to Use AI for University Study Without Breaking Integrity Rules (2026 Guide)

Most universities now allow some AI use, but the rules change from task to task. This guide shows how to find each assessment's AI rule, use AI as a study partner rather than a ghostwriter, reference it, and protect yourself if your work is questioned.

UniLearno Editorial Team

· 18 min read

Short answer: Most universities now allow some AI use, but the rules change from one assessment to the next. Check what each task allows, use AI to understand, plan and get feedback rather than to produce the work you submit, keep a record of how you used it, and acknowledge it the way your university asks. Last updated: 20 November 2026

Is it allowed to use AI at university in 2026?

Yes, in most places, but with conditions. The blanket bans of early 2023 have largely gone. In the UK, the 24 Russell Group universities agreed a set of principles back in July 2023 committing to help students and staff become "AI-literate" while upholding academic rigour and integrity. In Australia, the higher education regulator TEQSA has pushed universities to redesign assessment rather than rely on bans. Across the USA and Canada, policies are set by each institution and, very often, by each instructor in the course syllabus.

What this means in practice is that the answer to "am I allowed to use AI?" is almost always "it depends on the task". You might be encouraged to use AI to brainstorm in one unit, asked to use it and critique its output in another, and banned from using it at all in a third. A single semester can include all three.

Students are clearly using it. The Higher Education Policy Institute's Student Generative AI Survey 2026, based on 1,054 full-time UK undergraduates surveyed in December 2025, found that 95% of students use AI in some form and 94% use generative AI to help with assessed work. The same survey found that 12% had put AI-generated text directly into assessed work, up from 8% in 2025 and 3% in 2024. That last group is the one most at risk of a misconduct case.

The survey also found that students "still feel uncertain about what is acceptable". If that is you, you are in the majority, and the rest of this guide is designed to remove that uncertainty.

Why do universities care so much about how you use AI?

Because a degree is a promise. When a university awards you a qualification, it is telling employers, professional bodies and the public that you can do certain things: analyse a case, write a clear argument, run a statistical test, care for a patient safely. If an AI tool did the work, that promise is broken. That is why integrity policies focus on two questions:

  1. Whose thinking is this? Your submission should show your understanding, judgement and decisions.
  2. Has the reader been told the truth about how it was made? If a tool helped, the marker should know how, in the way the university asks you to disclose it.

AI use that keeps both answers honest is generally fine where it is permitted. AI use that hides who did the thinking is where the trouble starts. There is also a learning reason. If AI does the hard part, you skip the practice that builds skill. That catches up with you in exams, vivas, placements and, eventually, at work.

How do I find out what my assessment allows?

Treat every assessment as having its own AI rule, and look for it in this order:

  1. The assessment brief or task sheet. This is the most specific source and usually overrides everything else.
  2. The unit, module or course outline (or syllabus). Many outlines now include an AI statement for the whole unit.
  3. Your learning platform announcements (Canvas, Moodle, Blackboard and so on). Lecturers sometimes clarify rules mid-semester.
  4. Your university's academic integrity policy and AI guidance page. This sets the overall framework and the misconduct process.
  5. Your lecturer, tutor or unit coordinator. If the rule is unclear, ask in writing and keep the reply.

What do the common "AI levels" mean?

Many universities describe permitted AI use with a scale. One widely used model is the AI Assessment Scale developed by Mike Perkins, Jasper Roe and Leon Furze (revised version published in 2025). It has five levels:

  • Level 1, No AI: the task must be completed without AI, often under supervised conditions.
  • Level 2, AI-assisted planning: AI can help with brainstorming, outlining or early research, but the final work is yours.
  • Level 3, AI-assisted task completion: AI can help with parts of the task, such as feedback or editing, and you are usually expected to show or explain how you used it.
  • Level 4, Full AI: AI can be used throughout to complete the task, with you directing and evaluating it.
  • Level 5, AI exploration: you use AI creatively or critically, often as the subject of the task itself.

Your university may use different labels or numbers, but if your brief says something like "Level 2", it is very likely a version of this scale.

What is the "two-lane" approach?

Some Australian universities, led by the University of Sydney, sort assessments into two lanes:

  • Lane 1 (secure): assessments of learning, checked in person, such as exams, in-class tasks, vivas and oral assessments. Whether AI is allowed is decided by the examiner, and often it is not.
  • Lane 2 (open): assessments for and as learning, where AI use is "scaffolded and supported". You may be shown how to use AI to get feedback, suggest structures or summarise reading.

If your university uses lanes, the lane tells you a lot. Lane 1 tasks check what you can do on your own. Lane 2 tasks are where you are expected to learn to use AI well, and to be open about it.

What if the rule is unclear or there is no rule?

Do not assume silence means permission. A short, polite email works well:

"For the Week 8 report, may I use AI to quiz myself on the readings and to get feedback on my own draft? I won't use it to generate any text I submit." If the answer is yes, you have it in writing. If it is no, you have avoided a problem.

What counts as acceptable AI use, and what crosses the line?

Every task is different, but most university policies fall into a similar pattern. Use this as a guide, then check it against your own assessment rules.

Usually acceptable (where AI is permitted at all)

  • Asking AI to explain a concept you have met in a lecture, in simpler language or with a new example.
  • Generating practice questions and quizzing yourself on your own notes.
  • Talking through an idea to test your understanding, then checking what it says against your readings.
  • Getting feedback on a draft you wrote: clarity, structure, whether your argument follows.
  • Getting help to understand feedback from your marker.
  • Planning your study time or breaking a big task into steps.

Often allowed with acknowledgement (check the brief)

  • Brainstorming possible angles, then choosing and developing one yourself.
  • Asking for an outline structure as a starting point, then building your own plan.
  • Grammar and language suggestions on your own writing, particularly if English is an additional language. Note that some universities treat heavy rewording tools differently from spellcheck.
  • Using AI to help find search terms or sources, which you then locate, read and check yourself.

Usually misconduct (unless the task explicitly allows it)

  • Submitting AI-generated text, code, calculations or images as your own work.
  • Pasting AI output into your assignment and lightly editing it.
  • Using AI to paraphrase sources so they look like your own analysis.
  • Citing references an AI suggested without finding and reading them. AI tools can produce convincing references that do not exist.
  • Using AI in an exam, test or other secure assessment where it is not permitted.
  • Not disclosing AI use when the task requires you to.

A useful test: if you removed the AI tool, could you explain and defend every sentence of your submission in a conversation with your tutor? If not, something has crossed the line.

How do I use AI as a study partner rather than a ghostwriter?

The difference is who does the thinking. Here are five ways of working that keep the thinking with you.

1. Ask it to teach, not to answer

Instead of "What is the answer to question 3?", try:

  • "Explain the difference between fixed and variable costs as if I'm a first-year student, then ask me two questions to check I understood."
  • "I think opportunity cost means X. Is that right? Don't give me the full answer, just tell me what I'm missing."

You still do the work of understanding, but with a patient explainer.

2. Turn your notes into active recall

Paste in your own lecture notes (not copyrighted course materials if your university restricts that) and ask:

  • "Write ten short-answer questions from these notes. Don't show the answers until I've tried."
  • "Mark my answers and tell me which topics I should revise again."

Testing yourself is one of the most effective ways to learn, and AI makes it quick to set up.

3. Use it as a critic, never the author

Once you have written a draft, ask for feedback in specific areas:

  • "Where is my argument weakest in this paragraph?"
  • "Does each paragraph link back to the question? List any that drift."
  • "What counter-argument have I not addressed?"

Then you decide what to change, and you make the changes in your own words. Avoid "rewrite this better", which hands the writing to the tool.

4. Use it to plan your time

Big tasks are easier when they are broken down. Try: "I have a 2,000-word report due in three weeks and I work 15 hours a week. Help me plan the stages." You get a schedule; you still do every stage.

5. Check everything it says

AI tools can be confidently wrong. Treat every claim as something to verify in your readings, textbook or a library database. If a statistic, quote or reference matters, find the original. This is also where human help can make a difference. A tutor can tell you whether your understanding is right, which AI tools cannot reliably do, and can help you build the habits above. If you want that kind of support, you can find a subject expert for 1-on-1 guidance who will work through the process with you rather than doing it for you.

Worked example: using AI properly on a written assignment

This is an invented case to show the process. It is not a model answer.

The task: Maya is a second-year business student. Her assessment is a 2,000-word report on how a mid-sized retailer could make its supply chain more resilient. The brief says: "AI may be used for planning and feedback only (Level 2). Include a short AI use statement. AI-generated text must not appear in your submission."

Step 1: Read the rule and write it down. Maya copies the AI rule into the top of her notes document so she sees it every time she works. She notes three limits: planning, feedback, no AI text in the submission.

Step 2: Understand the topic herself first. She reads the two set readings and her lecture slides, and writes a half-page summary in her own words. Only then does she open an AI tool.

Step 3: Use AI to test her understanding. She asks: "Here's my summary of supply chain resilience. What important ideas have I missed or misunderstood? Don't rewrite it." The tool points out that she has not mentioned supplier diversification. She goes back to the readings to learn about it.

Step 4: Brainstorm, then choose. She asks for "five possible angles for a report on supply chain resilience for a mid-sized retailer". She rejects three as too general, combines two into her own angle (local supplier networks plus better demand forecasting) and writes down why she chose it.

Step 5: Plan in her own words. She drafts her own outline: introduction, current risks, two recommendations, implementation, conclusion. She asks the AI, "Does this structure answer the brief? What would a marker see as missing?" It suggests a section on costs and trade-offs. She agrees and adds it.

Step 6: Write the draft without AI. She writes every sentence herself, using her readings and three extra sources she found through the library database. She saves versions as she goes (Google Docs or Word version history does this automatically).

Step 7: Get feedback on her own draft. She pastes one section and asks: "Which sentences are unclear? Where do I make claims without evidence?" She gets four comments, accepts two, rejects two and rewrites the relevant sentences herself.

Step 8: Check references manually. Every source in her reference list is one she found, opened and read. She formats them using her university's referencing guide.

Step 9: Write the AI use statement. For example:

"I used [tool name] to check my understanding of the set readings, to brainstorm possible report angles and to get feedback on the clarity of my draft. All text in this report is my own. I did not use AI to generate any submitted text or references." Step 10: Keep her evidence. She keeps her notes, outline, version history and a copy of her AI chats in a folder until her final grades are confirmed. Notice what Maya did not do: she never asked the tool to write a paragraph, rewrite her draft or find her sources. The tool made her faster and more confident, but the report is entirely hers.

How do I acknowledge or reference AI use?

Follow your university's instructions first; they vary. Broadly, there are three ways universities ask you to show AI use:

  • An AI use statement or declaration, like Maya's above, explaining what tool you used and how.
  • A reference or citation, when you quote or rely on AI output directly and the task allows it.
  • An appendix with your prompts and the AI responses, which some tasks require.

How do I reference AI in APA 7th?

The APA Style team updated its guidance in September 2025. For a reference to a generative AI tool, the basic pattern is:

  • Format: Company name. (Year). Tool or model name [Description, e.g. Large language model]. URL
  • Example from APA's guidance: Anthropic. (2025). Claude 4 Sonnet [Large language model]. https://claude.ai/new
  • In-text: (Anthropic, 2025)

APA also describes a format for citing a specific chat, which adds the chat's title, date and "[Generative AI chat]" as the description. Check which version your university's library guide prefers.

What about Harvard and other styles?

Harvard has no single official version, so each university's library publishes its own rules for AI. Search your library website for "referencing AI" or "generative AI citation". Many ask you to cite AI as personal communication or as software, alongside an acknowledgement statement.

Is referencing AI enough to make it acceptable?

No. A reference shows where content came from; it does not give you permission to use it. If the task says no AI-generated text, citing it does not fix the breach. As APA's own guidance puts it, it is "still not a good idea to submit AI-generated work" as your own.

Can my university detect AI, and what if I'm wrongly accused?

AI detection tools exist, including features built into Turnitin, but they are not reliable enough to prove anything on their own. The TEQSA-commissioned report Enacting assessment reform in a time of artificial intelligence (September 2025) describes reliable detection as "all but impossible" and recommends that universities invest in assessment design rather than detection. In Canada, Mount Royal University's AI working group advised in August 2026 that staff avoid using AI detection applications in integrity processes, and summarised research showing high false-positive rates, especially for writers whose first language is not English. So universities increasingly rely on other evidence: your drafts, your version history, how your work compares with your earlier writing, and whether you can explain it in a conversation or viva.

How do I protect myself?

  • Keep your process. Write in a tool with version history. Keep your notes, outlines and reading summaries.
  • Keep an AI log for tasks where you used AI: what tool, what you asked, how you used the answer.
  • Know your sources. Be ready to say where every idea and reference came from.
  • Write in your own voice. Over-polished wording that does not sound like you can raise questions, even if you wrote it.

What should I do if I'm accused of using AI?

  • Stay calm and read the allegation carefully. Note what evidence is being used and the deadline to respond.
  • Read your university's misconduct procedure. It will explain your rights, including whether you can bring a support person.
  • Gather your evidence: drafts, version history, notes, AI log, search history in library databases.
  • Contact your student union, guild or advocacy service. Most universities have free, independent advice for exactly this situation.
  • Be honest. If you used AI in a way you now realise was not allowed, saying so early is usually treated better than denial.

What about exams, in-class tasks and oral assessments?

Expect more of them. TEQSA's 2025 guidance sets out pathways for universities to assure learning, including having secure tasks in each unit that work as hurdles you must pass. Secure tasks are completed under controlled conditions, often with identity checks, such as supervised exams, in-class writing, practical demonstrations and oral presentations or vivas.

For students, this changes the maths. If you lean on AI for your take-home work and never build the underlying skill, a secure task later in the unit can expose the gap. The best preparation for secure assessments is the study-partner approach above: use AI to practise and test yourself, so you can perform without it.

For oral assessments in particular, practise explaining your work out loud. If you can talk a friend or tutor through your reasoning, answer "why did you choose that?" and handle a follow-up question, you are well prepared.

Do the rules differ in Australia, the UK, the USA and Canada?

The principles are similar everywhere, but who sets the rules differs.

  • Australia: TEQSA, the national regulator, expects universities to show that students genuinely achieve learning outcomes. Many universities are adopting secure assessment approaches such as the two-lane model, with clear AI rules per task.
  • UK: There is no single national AI rule. Universities set their own policies, many guided by the Russell Group principles of July 2023. HEPI's 2026 survey recommends that universities publish clear, assessment-specific AI guidance, which suggests that rules will keep getting more specific.
  • USA: Policies are set by each institution and frequently by each instructor. Read your syllabus closely, because two classes at the same college may have opposite rules. Honour codes may also apply.
  • Canada: Similar to the USA, with institution-level policies and instructor discretion. Some universities, such as Mount Royal, have advised against relying on AI detection tools.

If you are an international student, remember that integrity cultures differ between countries. Something that was normal at school or in a previous university (for example, reusing published wording without quotation marks) may be treated as misconduct where you study now. When in doubt, ask.

How do I keep my own skills growing when AI makes things easy?

AI can make study feel faster, but speed is not the same as learning. A few habits help:

  • Attempt first, then ask. Try the problem, outline or paragraph yourself before you open an AI tool.
  • Explain it back. After the AI explains something, close the window and explain it in your own words.
  • Limit polishing. Let your writing sound like you. Markers value clear, honest writing over smooth, generic prose.
  • Build AI skills on purpose. Employers increasingly want graduates who can use AI critically. HEPI found that 68% of students see AI skills as essential, but only 48% said teaching staff were helping them build those skills. Learning to question, check and direct AI is a skill in itself.
  • Get human feedback too. Tutors, peers, writing centres and study skills advisers can spot things AI cannot, such as whether you really understood a concept.

FAQ

Can I use ChatGPT, Claude or Gemini to help with my assignment?

It depends on the assessment. Check the task brief and unit outline for the AI rule. Where AI is allowed, use it for understanding, planning and feedback, and acknowledge it as your university asks.

Is using Grammarly or a grammar checker cheating?

Basic spelling and grammar checks are usually fine. Tools that rewrite whole sentences or paragraphs may count as AI assistance, so check your university's guidance and disclose them if required.

Do I have to reference AI if I only used it to brainstorm?

Many universities ask for an AI use statement even for planning-only use. If the brief asks for disclosure, include a short statement saying how you used it.

Can AI detectors prove I used AI?

No. Detection tools produce false positives and are not reliable proof on their own. Universities usually look at other evidence, such as drafts and your ability to explain your work, which is why keeping your process matters.

What happens if I break the AI rules?

Outcomes range from a warning or resubmission to a zero for the task or more serious penalties for repeated or serious cases. Each university has its own procedure, and your student advocacy service can explain it.

Is it okay to put my lecture slides or readings into an AI tool?

Check your university's policy. Some restrict uploading course materials, copyrighted readings or other people's data into public AI tools. Your own notes are usually the safest material to use.

Get support that builds your skills

Using AI well is now part of studying well. The students who do best treat it like a study partner: something that explains, quizzes and challenges them, while the thinking and writing stay their own. If you would like a person to help you build those habits for a specific subject or assessment, see how UniLearno 1-on-1 sessions work. You can also browse more study guides on the UniLearno blog.

Sources

Found this useful? Share it:

Free · once or twice a week

Get new articles by email

Study skills, academic writing, research and student news, delivered in the evening your time. Unsubscribe any time.

RSS feed

Find my academic expert