The AI in higher education conversations I sit in on at board and academic board level come back to five questions: can we still trust our own awards, what does this do to our staffing model, what will it cost, what do students now expect from us, and who is liable when it goes wrong. Boards that turn those questions into decisions, and record them, satisfy TEQSA. Boards that hold the conversation and minute nothing do not.
This article sets out the questions I hear most often from governing bodies and academic boards, and how to structure the discussion so that it ends in a decision the minutes can show. It draws on fifteen years of TEQSA registration and governance work, and on the two years since the regulator asked every provider to show its plan.
Why AI in higher education conversations now have a regulator in the room
In June 2024 TEQSA issued a request for information requiring every provider to lodge, by 3 July 2024, a credible institutional action plan overseen by governance to address the risk generative AI poses to award integrity. The word that matters in that sentence is governance. TEQSA did not ask the learning and teaching unit for a plan. It asked the provider, and it expected the governing body to own the answer.
That changed the character of the conversation. Before 2024, AI was a topic the academic board discussed when someone raised it. Since then it has been a standing risk that Standard 5.2 on academic integrity and Standard 6.2 on corporate monitoring both reach, and at renewal TEQSA reads the minutes to see whether the plan the provider lodged was ever discussed again. I have written about what compliance staff should be assembling in using AI in higher education: what TEQSA compliance officers want to see.
Question one: can we still trust our own awards?
This is the question academic boards ask first, and it is the right one. It comes from Standard 1.4, which requires that assessment demonstrates achievement of learning outcomes, and from Standard 1.5, which requires that a qualification certifies what it claims to. If a student can pass every unit with a language model, the award no longer certifies anything.
The useful form of the conversation is not "should we ban it" but "at which points in each course do we need to be certain the work is the student's own, and how will we secure those points". TEQSA's 2023 assessment reform publication frames this as trustworthy judgment through multiple, contextualised approaches rather than detection. The decision a board can minute is a program-level map of secured assessment points, an owner, and a date for review.
Question two: what does this do to our staffing?
Corporate boards ask this one, usually as a cost question, and academic boards ask it as a workload question. Both are legitimate. Redesigning assessment across a course is real work, and marking secured assessments such as vivas and supervised tasks takes more academic time than marking essays did.
The conversation that goes nowhere is the abstract one about whether AI will replace lecturers. The one that produces a decision is narrower: which staff need development in assessment design, what that costs, whether the sessional budget for the next year reflects the new marking load, and how the academic board will know the work has been done. Standard 3.2 expects staff to be supported to teach well, and a board that has funded that support can show it.
Question three: what will it cost, and what do students expect?
Cost questions tend to arrive with a vendor proposal attached: a detection tool, a licensed model for students, a new platform. My advice to boards is to ask what problem the purchase solves against the Threshold Standards before approving it. TEQSA has not endorsed detection software as sufficient, and a board that has bought a tool and called that its plan will find the RFI response uncomfortable.
Student expectation is the newer strand. Students now ask whether they will be taught to use these tools, and employers are asking the same. The 2025 enacting assessment reform publication moves from principle to practice on exactly this point. The wider strategic questions are the subject of AI and the future of tertiary education, and the pedagogical ones of the role of AI in tertiary education.
Question four: who is liable when it goes wrong?
Directors ask this one privately, and they should ask it in the meeting. The scenarios are concrete: a graduate whose competence is challenged by a professional body, a student penalised on an unreliable detection score, personal data pushed into a public model by a staff member. Each is a governance failure with a different remedy.
The decision the board can record is a policy position on each: what evidence is required before an integrity finding, what tools staff may use with student data, and how the academic board reports integrity cases and their outcomes upward. Standard 2.3 on wellbeing and safety and Standard 7.3 on information management both bear on this, and an RFI will ask what the board decided.
Turning the conversation into minutes
The distinction TEQSA draws is between a board that talked about AI and a board that governed it. In my experience the minutes that satisfy an assessor record four things: the risk as the board understood it, the options considered, the decision and its owner, and the date the board will look at it again. A paper noted is not a decision made.
Structure the standing item that way and the AI in higher education conversations at your board stop being anxious and start being useful. The regulator's request in 2024 was for a credible plan overseen by governance. Two years on, the evidence it wants is that governance is still overseeing it.
Download the Darlo AI Governance Checklist
— a one-page guide to the decisions a governing body and academic board should have minuted on generative AI, 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
Does TEQSA require providers to have an AI policy?
TEQSA's June 2024 request for information required every provider to lodge a credible institutional action plan, overseen by governance, addressing the risk generative AI poses to award integrity. A policy alone is not what was asked for; a plan with governance oversight and evidence of implementation is.
Should the corporate board or the academic board own the AI conversation?
Both, with different questions. The academic board owns assessment integrity and course design under Standards 1.4, 5.2 and 6.3; the corporate board owns cost, staffing, liability and monitoring under Standards 6.1 and 6.2, and should receive the academic board's reports.
Is AI detection software enough to satisfy TEQSA?
No. TEQSA has not endorsed detection software as sufficient, and its assessment reform publications favour program-level assurance with secured assessment points over reliance on detection.
How often should AI be on the board agenda?
In my experience a standing item at the academic board and at least an annual report to the corporate board is the minimum that produces a usable record for renewal of registration.
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.