2026 is a significant year for AI governance in Australian higher education. The Tertiary Education Quality and Standards Agency has signalled that responsible use of artificial intelligence — particularly in assessment and academic integrity contexts — is a regulatory priority for the current cycle.

This is not a surprise. The speed at which AI tools have been adopted across the sector, and the genuine governance challenges that adoption has created, were always going to attract regulatory attention. What is less clear to many providers is what TEQSA's attention actually requires of them in concrete, demonstrable terms.

This post outlines the regulatory framework, the specific risk areas TEQSA is scrutinising, and what documentation and governance infrastructure higher education providers should have in place.

Why 2026 Is the Inflection Year

Australian higher education has been grappling with AI in assessment since generative language tools became accessible to students at scale in early 2023. Most providers responded initially through policy — acceptable use statements, assessment redesign guidance, academic integrity policy updates.

Policy, however, is not the same as governance. A policy document tells you what is permitted. Governance infrastructure demonstrates that what is permitted is actually being monitored, assessed, and enforced.

TEQSA's regulatory posture has evolved accordingly. The agency's 2025 review activity included assessment integrity as a theme, and the 2026 focus on AI-mediated risk reflects an expectation that providers have moved beyond policy statements to demonstrable governance arrangements.

The TEQSA Framework: Relevant Threshold Standards

TEQSA's regulatory authority derives from the Higher Education Standards Framework (Threshold Standards) 2021, administered under the Tertiary Education Quality and Standards Agency Act 2011.

Several Threshold Standards are directly relevant to AI governance in assessment contexts.

Domain 1: Student Participation and Attainment. Standard 1.4 requires that assessment is designed and implemented to ensure students demonstrate achievement of the required learning outcomes. AI tools that influence assessment outcomes — whether through generation assistance, automated grading, or adaptive learning — must be evaluated for their effect on the validity of that demonstration.

Domain 3: Teaching and Learning. Standard 3.2 requires that teaching and assessment are conducted by appropriately qualified staff exercising professional judgement. Where AI tools are used to support assessment processes, providers must be able to demonstrate that professional judgement remains the operative decision-making mechanism — not the AI output.

Domain 5: Institutional Quality Assurance. Standard 5.1 requires that providers implement systematic monitoring and evaluation of their educational activities. AI systems used in assessment are educational activities for the purposes of Standard 5.1. Their monitoring cannot be ad hoc; it must be systematic.

Domain 6: Governance and Accountability. Standard 6.1 requires that providers have governing bodies and management structures that provide effective oversight. Effective oversight of AI systems in assessment requires that governing bodies understand what systems are in use, what governance arrangements apply, and what the outcomes of those systems are.

Three Risk Areas TEQSA Is Scrutinising

Based on TEQSA's published guidance and the pattern of its 2025 review activity, three risk areas warrant particular attention.

Academic integrity. The use of AI tools in student assessment creates integrity risks at two levels: students using AI tools in ways that are not permitted by assessment design, and providers using AI tools in ways that may disadvantage or incorrectly assess students. Both risk levels require governance. The first requires clear policy, detection capability, and disciplinary frameworks. The second requires evaluation of the AI systems used, documentation of how they were validated, and monitoring of their ongoing performance.

Equitable access. AI tools in education are not neutral with respect to access. Students from backgrounds with less exposure to these tools, students with particular learning needs, and students for whom English is an additional language may be differentially affected by AI-mediated assessment approaches. Providers must be able to demonstrate that they have assessed these effects and that their AI use does not create systemic disadvantage.

Data governance for student information. AI systems in education process significant volumes of student data — assessment submissions, engagement data, learning analytics. The governance of that data — how it is used, who has access, how it is protected, what happens when a student requests deletion — must be documented and demonstrable. This connects to OAIC Privacy Act obligations as well as TEQSA's own expectations.

What "Responsible AI in Assessment" Requires in Documentation Terms

TEQSA's expectations for documentation are grounded in a logic of demonstrability: providers must be able to show, not merely assert, that governance arrangements are in place and operating.

For AI systems used in assessment, the minimum documentation set that any provider should be able to produce includes:

Policy documentation. A current policy governing the use of AI in assessment, covering what is permitted, what is prohibited, how compliance is monitored, and what consequences apply. The policy must be accessible to students and staff and must have been reviewed following any material change in the technology environment.

System registration. A register of AI systems used in assessment contexts, describing each system, its purpose, the data it processes, who is responsible for it, and the governance arrangements that apply.

Evaluation documentation. For each AI system, evidence of an assessment against relevant governance criteria: the system's effect on assessment validity, its equity implications, its data governance arrangements, and its alignment with institutional policy.

Monitoring records. Evidence that AI systems in use are being monitored on an ongoing basis — not assessed once and forgotten. Monitoring records should show who reviewed the system, when, what they found, and what was done in response.

Accountability records. Documentation of who is responsible for each AI system's governance, at what level of the organisation, and through what decision-making process.

Common Governance Gaps

Across the sector, several governance gaps appear consistently.

Systems registered nowhere. Many providers have deployed AI tools across faculties and schools without any central registry. When TEQSA asks what AI systems are in use for assessment, the answer involves a time-consuming survey rather than a maintained register.

No clear accountability. When AI tools are deployed through informal channels — a faculty member adopting a tool, a team using a product without institutional procurement — accountability for governance is unclear. Nobody owns the system's compliance posture.

No re-evaluation when tools are updated. AI tools change. Their underlying models retrain; their interfaces evolve; their terms of service change. A governance assessment conducted at tool adoption does not remain current without a process for re-evaluation when material changes occur.

Documentation that is aspirational rather than operational. Policy documents that describe how AI should be governed are not the same as evidence that governance is occurring. TEQSA is looking for the latter.

Cross-Reference: NHMRC Research Ethics

For universities with significant research activities, AI governance obligations extend beyond teaching and assessment into research ethics. The National Health and Medical Research Council's National Statement on Ethical Conduct in Human Research (2023 update) addresses AI-assisted research activities, including the use of AI in data analysis, decision-support, and participant interaction.

AI systems used in NHMRC-funded research, or in research involving human participants, require ethics committee consideration that addresses the specific capabilities and limitations of the AI tools involved. Generic ethics approvals that do not address AI-specific risks are unlikely to be adequate where those risks are material to the research design.

Providers should ensure that their research ethics infrastructure has a mechanism for identifying when AI tools are being used in research contexts and for requiring appropriate ethics consideration of those tools.

How Governance Infrastructure Helps

The documentation and monitoring demands that TEQSA's expectations create — and that NHMRC adds to for research-active institutions — are not easily met through manual, ad hoc processes at sector scale.

A governance platform provides the structural infrastructure: an AI system registry where each tool is registered with its purpose, risk assessment, and accountability assignment; an evidence hub where governance documentation is stored with integrity verification; a control evaluation pipeline that assesses each system against applicable framework requirements; and a TrustScore™ — a deterministic composite governance score — that provides a board-reportable confidence signal.

The TrustScore is calculated from control decisions and verified evidence, not from AI model outputs. It reflects documented current state. When TEQSA asks for evidence of governance, the audit trail, the evidence record, and the score history are already assembled.

Download the AI Governance Checklist to assess your institution's AI governance posture. For a broader look at continuous assurance infrastructure, see Continuous Assurance vs Point-in-Time Audits. To see the platform for universities, visit Sentrify for Universities.

This post is informational and does not constitute legal or regulatory advice. TEQSA's regulatory expectations are grounded in the Higher Education Standards Framework (Threshold Standards) 2021 and TEQSA's published guidance. Higher education providers should seek advice from qualified legal and regulatory specialists when determining their specific obligations. References to NHMRC guidance are based on the National Statement on Ethical Conduct in Human Research (updated 2023) and relevant NHMRC publications.