AI literacy in tertiary education is built through dialogue rather than policy alone: a provider has to tell students plainly what is permitted in each unit, explain why, require disclosure of what tools were used, teach the judgment needed to use those tools well, and give staff the same clarity and the same training before asking them to enforce anything. A policy that students discover only when they are accused of breaching it has failed as communication, and TEQSA reads that failure as an academic integrity problem rather than a student one.
This article is about the conversation, not the rulebook. It draws on fifteen years of TEQSA registration and compliance work with private providers, some of which talked with their students about AI early and some of which did not.
Why TEQSA cares how providers talk about AI
TEQSA's 2023 paper on assessment reform for the age of artificial intelligence set two principles, and the first is that assessment and learning experiences must equip students to participate ethically and actively in a society pervaded by AI. That is a statement about literacy, and it means a provider cannot discharge its obligations by prohibition.
The Threshold Standards supply the rest. Standard 7.2 requires that students receive accurate, timely information about their obligations and the provider's expectations. Standard 1.3 requires orientation that prepares students for their studies, and Standard 5.2 requires that students and staff be informed about academic integrity expectations and their responsibilities. Read together, those standards make the dialogue about AI a compliance matter, and in my experience assessors now ask to see how it happens rather than whether a policy exists.
Expectations and permitted use, unit by unit
The most useful thing a provider can do is abandon the idea of a single institutional rule. What is permitted in a first-year writing unit differs from what is permitted in a final-year capstone, and students know it. The workable approach, and the one I see spreading across the sector, is a short statement in every unit outline that says what AI use is permitted for each assessment task, what is not, and what must be disclosed.
The conversation then happens in the first week of each unit, when the teacher walks students through the statement and explains the reason for it. That explanation is where AI literacy tertiary education actually begins, because a student who understands that a task is secured so that the provider can vouch for the award is being treated as an adult, and a student who is simply told no is being invited to test the boundary. Our piece on conversations on AI in higher education sets out how providers are structuring those first-week discussions.
Disclosure as a habit, not a confession
Disclosure works when it is routine and consequence-free for permitted use. A one-line acknowledgment attached to every submission, stating which tools were used and for what, normalises honesty and gives markers useful information. Where a provider treats disclosure as an admission that invites suspicion, students stop disclosing, and the provider loses the only visibility it had.
Markers need to know what to do with a disclosure, how to distinguish permitted use from misconduct, and when to refer. In my experience the providers whose integrity processes collapsed under AI were not those with weak policies but those whose staff had never been told how the policy applied to a real submission, and who therefore referred everything or nothing.
Teaching the AI literacy tertiary education now requires inside the curriculum
Literacy cannot be delivered by a policy page or a single orientation module. It has to be taught, assessed and revisited, which means the academic board needs to decide where in each course students learn to use AI tools, evaluate their output, understand their limits and recognise when their use is inappropriate. TEQSA's 2025 paper on enacting assessment reform takes this from principle to practice, favouring program-level design in which some assessment points are secured and others deliberately require students to work with AI and account for how they did.
For a private provider with a small academic team, the practical route is to map the course once: which units teach AI use explicitly, which assess it, which are secured, and how the sequence develops across years. That map is the AI literacy tertiary education providers can show an assessor, and it is a far stronger answer to TEQSA's questions than a list of prohibitions.
Bringing staff into the conversation
Staff are often more anxious about AI than students, and the dialogue has to include them. Teaching staff need training in the tools, in redesigning assessment, in reading a disclosure and in the misconduct process. Sessional staff, who mark most of the work at many private providers, need it most and usually receive it least.
The policy conversation should also be two-way. Teachers see how students actually use these tools long before a policy committee does, and a provider that collects that observation through unit reviews and the academic board's integrity reporting keeps its position current. In my experience the policy written once and never revisited is the one assessors find least convincing. Where the sector's thinking is heading is the subject of our article on AI and the future of tertiary education.
What I tell providers about the AI conversation
Say what is permitted, task by task, and say why. Make disclosure ordinary, and teach the literacy as part of the course rather than as a warning at the start of it. Train the staff before you ask them to police anything, and let what they see feed back into the policy. A provider that does those things has a dialogue TEQSA can observe in unit outlines, orientation records, staff development logs and academic board minutes. Our guide to generative AI in higher education collects the regulator's publications the conversation should draw on.
Download the AI Governance Checklist
— including a student and staff communication plan for permitted use, disclosure and literacy, drawn from our TEQSA registration and governance work with private providers. Get the checklist
Want the full article?
Enter your email for free access to the rest of this guide and our TEQSA resource library.
Frequently asked questions
Should a provider ban generative AI for students?
No. TEQSA's assessment reform principles expect providers to equip students for a society pervaded by AI, so the workable approach is to specify permitted use task by task, secure the assessments that vouch for the award, and teach students to use the tools well.
How should students disclose their use of AI?
Through a routine acknowledgment attached to each submission stating which tools were used and for what. Disclosure works only when permitted use carries no penalty, so that students have no reason to conceal it.
Where does AI literacy belong in the curriculum?
Mapped across the course, with specific units that teach AI use, assess it and secure the points that verify learning. The academic board should approve that map, and it becomes the evidence TEQSA sees.
What training do teaching staff need on AI?
In the tools themselves, in redesigning assessment, in interpreting a student's disclosure and in the provider's misconduct process. Sessional markers need this training most and should not be left out.
Dr Brendan Moloney is CEO of Darlo Higher Education, Australia's largest specialist TEQSA consultancy. He holds a PhD from the University of Melbourne, is a Cambridge University Press author on governance in higher education, and has advised private providers on registration and course accreditation for more than fifteen years.