AI and assessment in VET: what ASQA’s principles actually ask of RTOs

Generative AI has made one question urgent for every RTO: how do you know an assessment shows the student’s own skills? ASQA has answered with a set of principles for the responsible use of AI in VET. They do not add new rules. Instead, they show how your existing obligations under the 2025 Standards apply when AI is in the picture.

ASQA’s five principles

  1. Governance: clear accountability, policies and controls for how AI is chosen and used.
  2. Human oversight and accountability: people stay responsible. ASQA says decisions affecting students “remain the responsibility of qualified trainers, assessors and staff”.
  3. Privacy and data security: handle student information securely, including inside AI tools.
  4. Equity, inclusivity and wellbeing: AI use should not disadvantage any group of learners.
  5. Alignment with training products and industry: AI use must fit the training package and real workplace practice.

What it means for assessment

On assessment, ASQA’s guidance is clear on four points:

  • Assessment outcomes must reflect the student’s own skills, knowledge and competencies.
  • Assessment judgements must not rest solely on AI. An AI tool can help, but a qualified assessor makes the call.
  • Providers should monitor for AI misuse.
  • Providers should identify which existing assessment tasks can no longer be relied on now that AI can complete them.

Two examples worth copying

Case studies on ASQA’s website show the practical direction:

  • In one, a written task flagged as possibly AI-generated was resolved by asking the student to explain their work in a recorded video, with their identity verified.
  • In another, an assessment was redesigned so students had to explain the reasons behind their decisions, not just produce an answer.

The common thread: assess the thinking, not just the output. Observation, verbal questioning and workplace evidence are much harder to fake than a written response.

Higher education is heading the same way

TEQSA, the higher education regulator, published guiding principles on “assessment reform for the age of AI” in November 2023. In September 2025 it followed with “Enacting assessment reform in a time of artificial intelligence”, which argues for redesigning assessment rather than relying on AI detection, which it describes as not viable.

A self-check for your RTO

ASQA’s principles map to several standards, including 1.1 to 1.4 on training and assessment. Practical questions to ask:

  • Who is accountable for AI use in training and assessment?
  • Do your LMS or SMS have built-in AI features, and where is that data stored, onshore or offshore?
  • Do staff know what they must never paste into an AI tool?
  • Do students know which AI tools you use, and have you got prior written consent before using their personal or sensitive information?
  • Are AI-generated records kept like any other assessment record?
  • Which of your assessment tasks could a chatbot complete today?

A note on ASQA’s own AI use

ASQA practises what it preaches. Its AI transparency statement (updated March 2026) says it uses AI for workplace productivity and analytics, but not in public-facing services, regulatory functions or decision-making, and that humans make final decisions.

This article is general information, not compliance advice.

Sources

Scroll to Top