Here is the trap almost nobody warns you about. You sit down for an open-book exam, your notes are open, a browser tab is open, and somewhere in the back of your mind a voice says: “It is open-book, so surely I can just ask ChatGPT.” That single assumption has ended up in front of academic misconduct panels more often than most students realise.
I want to close that gap directly, because when I went looking for a straight answer to this exact question, every top result answered a broader one instead. They explain AI policy for assessment in general, but they go quiet the moment you are actually sitting at the desk with the clock running. So let me take you through what “open-book” really permits, where AI sits in that picture, and the one habit that keeps you safe.
What “open-book” actually means
Open-book is a narrower permission than it sounds. In most Australian universities, an open-book exam simply means you are allowed to bring and consult certain reference materials during the exam, usually your own notes, prescribed textbooks, and sometimes printed readings. The University of Melbourne describes open-book exams as assessments where you can refer to resources, but stresses that they still test your understanding rather than your ability to look things up (University of Melbourne, n.d.).
The key word there is “certain”. Open-book does not mean “anything goes”. It means a specific, defined set of materials is permitted, and everything outside that set is still off limits unless the exam paper says otherwise. UNSW makes a similar point in its guidance on open-book and take-home exams, noting that these formats reward how well you can apply and organise knowledge under time pressure, not how much you can find on the day (UNSW Sydney, n.d.).
So the first thing to understand is that “open-book” and “AI-allowed” are two completely separate questions. One being true tells you nothing about the other.
Why AI is a separate permission, not an extension of open-book
This is the part that catches people out. It feels intuitive that if you can consult your notes and the internet, you can consult an AI tool. But universities treat generative AI as its own category, governed by its own rules, precisely because it does something notes and textbooks cannot: it can generate answers for you rather than help you recall your own.
That distinction matters because the whole point of an exam is to measure what you can do. TEQSA, the national regulator for higher education, is blunt about this in its advice for students. It states that AI use may be restricted or banned depending on your university’s policies, that the rules differ between disciplines and even between individual assessment tasks, and that using AI in a way that is inconsistent with your institution’s rules counts as academic misconduct (TEQSA, n.d.). The regulator’s own words are worth sitting with: you need to understand the expectations for each assessment task, not just each subject or institution.
Read that again, because it is the crux of this whole article. The permission to use AI attaches to the specific task, not to the format label. An open-book exam and a take-home essay in the same subject can have opposite rules on AI, and both can differ from your university’s general position.
The rule that actually governs your exam: the paper itself
So where do you find the answer? Not in a blanket university policy, and not by reasoning from the word “open-book”. You find it on the exam paper, the cover sheet, or the assessment instructions in your learning management system (the online system your uni uses for subjects, such as Canvas, Blackboard, or Moodle).
The University of Queensland’s rules for using AI in assessment set the general expectation clearly: students must only use AI in the ways their course and task instructions permit, and the responsibility sits with the student to check (The University of Queensland, n.d.). What that guidance does not do, and what almost no institutional page does, is spell out the exam-specific scenario: you, mid-exam, with a browser open, wondering whether opening an AI tool is allowed.
That silence is exactly why the safe default matters. If the exam instructions do not explicitly permit AI use, treat it as not permitted. Absence of a prohibition is not permission. In an exam context, the burden is on you to confirm that a tool is allowed, not on the university to list every tool that is banned.
Invigilated and lockdown-browser exams: the mechanics that block AI
There is also a practical layer here, and it changes depending on how your open-book exam is run.
Some open-book exams are invigilated, meaning they are supervised, either in a physical exam hall or remotely through monitoring software. Many are also run through a lockdown browser, which is software that restricts your computer to the exam environment and blocks access to other applications, tabs, and sometimes the wider internet for the duration.
Here is the important nuance. An exam can be described as “open-book” and still run on a lockdown browser that blocks AI tools entirely. “Open-book” refers to what reference materials you are permitted to consult; the lockdown browser is a technical control over what your machine can access. The two are not in conflict. A university can say “you may use your own notes” while the software simultaneously prevents you from opening a browser tab to an AI chatbot at all.
This is very different from a take-home open-book assessment, where you have your own device, your own internet connection, and no software restriction. In a take-home task, the only thing standing between you and an AI tool is the rule on the paper. That makes reading the instructions even more important, because there is no lockdown browser doing the enforcing for you. The temptation is higher and the technical barrier is lower, so the rule is the only thing protecting you.
What happens if you use AI when it was not permitted
Let me be direct about the consequences, because the “grey area” framing does students a real disservice.
Using an AI tool during an exam when it has not been explicitly permitted is not usually treated as a grey area. It is typically treated as academic misconduct, in the same bracket as bringing unauthorised materials into any exam. TEQSA positions the misuse of generative AI alongside contract cheating in its academic misconduct resources, which tells you how seriously the sector takes it (TEQSA, n.d.).
The process at most Australian universities follows a broadly similar shape. An allegation is raised, usually by the marker or the exam supervisor. You are notified and given a chance to respond, often in writing and sometimes at a meeting or hearing. A decision is made on the balance of probabilities, and if misconduct is found, penalties can range from a mark of zero on the task, to failing the subject, to suspension in serious or repeat cases. The University of Sydney’s academic integrity guidance on artificial intelligence sets out this kind of framework, making clear that unauthorised AI use in assessment can trigger the misconduct process (The University of Sydney, n.d.).
The reason I want you to sit with this is not to frighten you. It is because the anxiety of “did I just cross a line” is entirely avoidable with about ninety seconds of reading before the exam starts.
Your exam-day checklist
Here is what I do, and what I would tell any mate a year behind me in the degree to do.
First, read the exam cover sheet and the LMS assessment instructions before the exam, not during it. Look specifically for the words “artificial intelligence”, “AI”, “generative”, or the name of a tool. If the paper lists permitted materials, check whether AI appears on that list.
Second, if the AI position is unclear, ask the subject coordinator directly, in writing, before exam day. A short email saying “Can you confirm whether any AI tools are permitted during the open-book exam for this subject?” gives you a written answer you can rely on. Do not ask a classmate and treat their guess as fact.
Third, never assume a general subject AI policy carries over into a timed, invigilated exam. A subject that lets you use AI to brainstorm an essay draft may prohibit it entirely in the exam. The task-level rule always wins.
Fourth, if AI is genuinely not addressed anywhere and you cannot get an answer in time, do not use it. The safe default protects you.
If you want the broader picture of how different institutions frame these rules, I have written a companion guide on Australian university AI policies that maps the general landscape your exam rule sits inside.
If AI is explicitly permitted: declare it properly
Sometimes the answer is yes. Some open-book assessments, particularly take-home ones, do permit AI use, occasionally they even require it. If that is your situation, the job is not finished once you have used the tool. Where AI use is allowed, you almost always have to acknowledge it, and doing that correctly is part of staying compliant.
TEQSA reinforces this: where AI use is permitted, students need to understand how to reference it (TEQSA, n.d.). The mechanics of declaring AI use in an exam mirror the process for assignments closely, so rather than repeat it here, I would point you to my guide on citing AI use in university assignments, which walks through what a proper declaration looks like and how to frame it across the common referencing styles.
The principle to hold onto is simple. Permission plus honest declaration is safe. Silence, guessing, or assuming is where students get into trouble.
Walking in already knowing
The real problem here is timing. Exam day is precisely when you have the least capacity to research a policy question calmly, and precisely when the pressure to “just check one thing” is highest. The fix is to move that decision earlier, into your revision period, when you can read the instructions properly and email a coordinator without a clock running.
That is the mindset GradeMap is built around. I am building it because I needed it: a way to plan exam-period workload and revision so that by the time you walk into an open-book exam, you already know what is allowed instead of guessing under pressure. Knowing the rule before you sit down is not a small thing. It is the difference between a calm exam and a misconduct meeting you never saw coming.
References
The University of Queensland. (n.d.). Rules for using AI in assessment. Institute for Teaching and Learning Innovation. https://itali.uq.edu.au/teaching-guidance/generative-ai-teaching-learning-and-assessment/rules-using-ai-assessment
The University of Sydney. (n.d.). Artificial intelligence. Academic integrity. https://www.sydney.edu.au/students/academic-integrity/artificial-intelligence.html
Tertiary Education Quality and Standards Agency. (n.d.). Artificial intelligence: Advice for students. https://www.teqsa.gov.au/students/artificial-intelligence-advice-students
Tertiary Education Quality and Standards Agency. (n.d.). Student academic misconduct resources. https://www.teqsa.gov.au/students/student-academic-misconduct-resources
University of Melbourne. (n.d.). Open-book exams. Academic Skills. https://students.unimelb.edu.au/academic-skills/resources/study-skills/exam-preparation/open-book-exams
UNSW Sydney. (n.d.). Open-book and take-home exams. Current Students. https://www.unsw.edu.au/student/managing-your-studies/academic-life/assessments-exams-resources/exams/preparation/open-book
FAQ
Does an open-book exam automatically allow AI tools like ChatGPT?
No. Open-book means you may consult a defined set of reference materials, usually your own notes and set texts, and sometimes the internet. AI use is governed separately by the specific exam instructions. If the paper does not explicitly permit AI, treat it as not allowed, because absence of a ban is not the same as permission.
Where do I find out whether AI is allowed in my exam?
On the exam paper itself, the cover sheet, or the assessment instructions in your learning management system, not in a general university policy. The permission attaches to the individual task. If it is unclear, email your subject coordinator before exam day and get the answer in writing.
Will a lockdown browser stop me from using AI in an open-book exam?
Often, yes. A lockdown browser restricts your computer to the exam environment and can block other tabs, apps, and the wider internet, including AI tools, even when the exam is otherwise open-book. This is different from a take-home open-book assessment, where there is no software control and the rule on the paper is the only thing enforcing the boundary.
What happens if I use AI in an exam when it was not permitted?
It is typically treated as academic misconduct rather than a grey area. The usual process involves an allegation, a chance for you to respond, and a decision that can lead to a zero on the task, failing the subject, or suspension in serious cases. Australia’s regulator groups generative AI misuse alongside contract cheating, which reflects how seriously it is taken.
If AI is allowed in my open-book exam, do I still need to declare it?
Yes. Where AI use is permitted, you almost always have to acknowledge how you used it, and failing to declare it can itself become an integrity issue. The declaration process mirrors the one for assignments, so follow your subject’s required referencing approach and keep a note of which tool you used and how.