Responsible AI in VET Starts with Governance

ASQA’s August 2026 guidance gives RTOs a clear structure for responsible AI use. Its five principles do not introduce new regulatory requirements. They help providers interpret, implement and oversee AI within existing obligations, including the 2025 Standards for RTOs.

An RTO can govern AI through its existing responsibilities for training and assessment, student support, workforce capability, information management and organisational accountability. Each AI use case should connect to those responsibilities.

For CEOs and governing persons, the practical question is whether the organisation can explain where AI is used, who owns each use, how risks are controlled and what evidence shows the controls are working.

The five principles as a governance test

ASQA’s principles cover five connected areas.

• AI use is supported by strong governance that ensures it does not undermine the quality or integrity of VET.
• Human oversight and accountability are maintained in all AI supported activities, ensuring that decisions affecting students remain the responsibility of qualified trainers, assessors and staff.
• AI systems and tools manage information securely and in accordance with existing privacy, data protection and record keeping obligations.
• AI use supports and enhances student equity, inclusivity, accessibility and wellbeing.
• AI use aligns with training product requirements, industry expectations and the needs of the relevant student cohort.

These principles apply across the full student journey. An AI tool used to draft learning material creates different risks from one used to recommend student support, analyse assessment evidence or prepare a response to ASQA. Governance should be proportionate to the purpose and potential impact of each use.

Build an AI use register

Start by recording the AI already present in the organisation. This includes stand-alone tools and AI features embedded in student management, learning management, marketing, support and assessment systems.

For each approved use case, record the following.

• its purpose and intended users
• the information entered, generated, stored or shared
• the training products, processes and student cohorts affected
• the operational owner and approving role
• required human review and escalation points
• known risks, controls and evidence
• the next review date and events that trigger an earlier review

The CEO and governing persons set the organisation’s risk settings and reporting expectations. An operational owner remains accountable for each use case. Privacy, security and records specialists assess information handling. Trainers and assessors decide whether a tool is suitable for training and assessment and whether its outputs support valid judgements.

Keep human accountability visible

ASQA is direct about student-impacting decisions. Qualified trainers, assessors and staff retain responsibility. AI can draft, recommend, classify or flag. The accountable person must review the output in context and own the decision.

RTOs should identify every AI-supported point that may affect assessment, progression, access, support or intervention. The process should state who reviews the output, what source evidence they consider, when an override is required and how uncertainty is escalated. Decision records should show the person responsible and the basis for the outcome.

Staff also need enough practical capability to recognise inaccurate, incomplete, biased or unsuitable outputs. A policy acknowledgement alone gives little assurance. Scenario testing, observed practice and review of actual decision records provide stronger evidence that human oversight is active.

Test the use case against competency and cohort needs

ASQA’s August newsletter described a provider testing an AI avatar for assessment in BSBPEF502 Develop and use emotional intelligence. Reviewers found that students could use predictable trigger phrases to gain positive responses. Some participants with accents had responses misinterpreted or not recognised by the speech-recognition system. The avatar did not reliably detect body language, eye contact or active listening behaviours.

The tool could not, by itself, show that a student could perform effectively with people. The provider retained the avatar for training and practice, while assessment used assessor observation, professional discussion and role play involving real people.

Before implementing the new tool, the provider checked the unit’s performance criteria, assessment requirements and assessment conditions, then tested it with assessors, industry representatives and instructional designers. That process exposed weaknesses in validity, fairness and sufficiency.

Apply the same discipline to other AI use cases. Map the tool to the relevant requirements. Test it with the intended cohort, including varied accents, abilities and levels of digital confidence. Record where it performs poorly and provide suitable alternatives. Repeat the review when the product, model, training product or cohort changes.

Control information at the point of use

AI governance must reach the screen where staff use the tool. People need clear rules for the information they can enter, approved systems, retention requirements and the handling of generated records.

Before approving a tool, establish how it collects, uses, stores and shares information. Check data location, access controls, vendor terms, deletion arrangements and any privacy notices or consent requirements. Record the assessment and the approval decision. An off-the-shelf product still requires provider due diligence.

Make evidence match practice

ASQA has observed generic wording that is not contextualised in provider applications and responses; submissions containing visible AI prompts or drafting; examples or practices not supported by the evidence provided; and corrective-action responses that may not genuinely reflect the provider’s actual practices, improvements made or evidence of implementation.

Treat every AI-supported submission as an evidence-control process. A named reviewer should check each statement against current operations and the attached evidence. Corrective actions should identify what changed, who owns the change, when it was implemented, how effectiveness will be checked and where the supporting record sits.

The approved version, reviewer, source material and supporting evidence should remain connected. That trail helps the RTO demonstrate an authentic account of its systems and gives leaders confidence that a polished document reflects real work.

Give leadership a review view

AI oversight belongs in the RTO’s existing governance cycle. Reviews should consider performance, incidents, staff overrides, student feedback, accessibility issues, data handling and continued alignment with training products and industry practice.

Leadership reporting should show approved use cases, accountable owners, higher-risk activities, upcoming and overdue reviews, incidents, corrective actions and unresolved exceptions. Governing persons can then see whether controls remain current and whether student-impacting decisions continue to receive qualified human oversight.

A practical starting sequence

Catalogue current AI use. Assign an owner to each use case. Identify student-impacting decisions and required human sign-off. Test higher-risk uses against training product and cohort needs. Document information controls. Schedule review dates and define the evidence leadership will receive.

Octossure gives RTOs one place to assign ownership, schedule reviews, retain evidence, track corrective actions and give leaders visibility of AI-related risk. If you are building AI governance into your quality and compliance systems, book a demo to see how this can work in practice.

Official ASQA sources

These official ASQA sources informed this article.

• ASQA IQ, August 2026: app4.vision6.com.au/em/message/email/view.php
• ASQA, Responsible use of Artificial Intelligence (AI) in VET: asqa.gov.au/for-providers/guidance-and-resou…
• ASQA, Artificial Intelligence (AI) case studies: asqa.gov.au/for-providers/guidance-and-resou…