The AI Mistakes ASQA Is Flagging in RTO Submissions (Plus a Free Tool to Check Yours)
ASQA's August 2026 edition of ASQA IQ was devoted entirely to AI. Whilst confirming that the 5 principles for the responsible use of AI in VET are not regulatory requirements, ASQA did outline some of the outputs being seen across the sector in relation to provider submissions. Rookie mistakes, and it’s obvious that Professional Development in, and by, RTO governance is a must.
Four patterns stood out to the regulator when reviewing provider submissions in response to regulatory actions and applications.
Generic wording
Applications and responses that describe a process: lots of words explaining a process, a little abstract, but not actually demonstrating how the RTO carries out the process. There is no context, just words. It reads competently, but really, when you focus, it doesn’t say much.
AI prompts remaining
Ridiculous, and a rookie mistake. ASQA's own example is almost funny if it were not so common: a submission that included the line "Would you like me to draft a formalised report on this?" Left in from the copy/paste and sent to the regulator! It is the clearest possible evidence that nobody read the final version before it went out.
Unsubstantiated claims
Responses that reference examples or practices that supporting documents don’t substantiate. These claims of compliance cannot be verified, which suggests the AI-generated content had little or no human review and may be wholly or partly inaccurate.
Inaccurate corrective actions
A response to a non-compliance finding where the response doesn’t reflect what the provider actually did. A response that is simply an ideal, not an action that took place in real life. Not an action that can be evidenced.
These instances are just examples of what the sector, and the world, are seeing in amateur AI use. From ASQA’s perspective, examples like these raise suspicion and doubt as to an RTO’s capability to maintain compliance. In my opinion, instances like these question the user’s integrity, intent and ability to fully understand, and/or articulate, their operational functions. Hence why ASQA developed the 5 principlesfor the responsible use of AI in VET. To help providers use AI responsibly.
The principles
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.
We see far too often across the sector, a misuse of AI technology, and not just from RTOs, but from consultants, system developers, resource writers, it’s everywhere. How many examples can you spot across LinkedIn right now where AI has been used in error? Where a colour coding suggestion has been left in a marketing post? Or where an AI generated image shows a person with 6 fingers? Or where a reference to a clause, implementation date or regulatory requirement is incorrect?
Too many.
So, what is the solution? Well, the National Regulator has taken to provide these 5 principles as guidance to support the sector to think deeper about their AI use and where it appears across our operations. It aims to improve AI use and result in higher quality outcomes across the board. So, of course, I took these 5 principles, and the guidance materials ASQA provided and put them into our AI Claude (which we affectionately refer to as “Barry”, our AI VET Compliance and RTO Administration Assistant) that we have established (quite extensively and very well I might add) and asked it to analyse ASQA’s position and review our Quality Management System policies, procedures and other tools, then tell me what needs to be updated.
After a fair bit of work preparing the project, Barry provided me with a summary of ASQA’s guidance, and RTO Implementation Guide and a spreadsheet tracker listing over 100 documents to work through (inclusive of drop-down items to track progress, conditional formatting to cross out completed works, a tiered approach and priority sorting and some other rubbish that really was just overkill). But nonetheless, the work I would have done in analysing and preparing the documents to update would have taken me hours… Barry did it in about 10 minutes or less.
From this, we reviewed the materials, scrapped about 10 documents that were not needed, brainstormed a couple of PD courses to create, updated Barry’s memory and got to work… one document at a time… With Barry’s help, of course. Updating our entire QMS suite with changes to support the 5 principles took me and Barry about 18 hours.
As I worked through our materials, I began to realise how embedded AI has become into our daily operations, how much I personally rely on it to be more efficient and organised. Barry acts as my PA type tool that keeps me organised, gives me tips, and lets me brainstorm concepts and strategy ideas. I use Barry to:
Prepare digital marketing strategies and content to allow CWTS to retain a presence in an increasingly competitive environment
Develop tools, documents, policies and procedures to a set format
Develop PD course content at a much faster rate
Research sector changes, trends and offerings and provide reference links so I can be sure
Enhance the presentation of our courseware through AI in design software and coding
Enhance the presentation of our digital media creatives through AI design software
Help us plan and prepare for meetings, training sessions, webinars and events
Transcribe recorded meetings to generate meeting minutes
Prepare and monitor our "to-do" lists through our Teamwork system integration and task tracking
Help me with my fortnightly financial analysis (directly through Xero integration)
Identify outdated documentation, products and other materials (directly through Microsoft SharePoint and OneDrive integration)
Analyse, summarise, screen and review my inbox to organise my time better
Prepare and send emails to improve efficiency
Search for things in my email inbox and through our Sharepoint system
Develop tracking sheets, spreadsheets and include formulas and formatting
Develop content in all sorts of places, including blog posts like this
Proofread completed works and identify areas for improvement
And most likely so much more that I can’t think of right now
And I am very, VERY adamant that none of this work is done without human oversight. Our CWTS Code of Conduct and Professional rules mandate it.
Yet some RTOs are using it for human functions or are lacking oversight and rulings as to how it should be used. For example, RTOs are using AI to:
Mark assessments and accepting the recommended outcome
Validate assessment tools and recording the coverage percentage as the validation outcome
Map an assessment tool to a unit and filing the mapping without a human checking it
Write assessment tools and putting them into use unverified
Write learner resources that don’t align to the unit or aren’t detailed enough
Flag a student’s use of AI and recording NYC on that basis alone
Write feedback, so every student gets the same generic comments
Determine an RPL outcome from a candidate's portfolio
Develop a TAS, which then describes delivery that the RTO doesn't actually perform
Generate marketing copy that does not comply with the Standards
Transcribe a student support conversation without asking the student
Get a second opinion by pasting a student's assessment evidence into the tool
Build a report or spreadsheet by pasting student records into the tool
Act as an avatar for a role play assessment where the unit needs real human interaction
An assessment decision has to be valid and reliable, whether a person, a spreadsheet or a language model helped produce it. Student information must be handled lawfully, whether it sits in a filing cabinet or gets pasted into a chatbot. Marketing content needs to be accurate and informative, whether a person or a model wrote it. RTO obligations don’t change here, even if technological processes do.
It is now up to you, the RTO, to work out where AI is used within your RTO and which of your existing obligations now touch on the use of AI that you may not have fully accounted for, and make sure the practice still meets the Standards. And after having just done it with our QMS tools, AI-related processes will be scattered across your software systems, privacy obligations, assessment rules, marketing requirements, and record-keeping practices you already have. It will be embedded into your systems, whether you enabled them or not, and your staff are likely using AI tools in ways you never even knew.
Ask yourself:
Do you know whether your student management system has AI features switched on by a vendor update, without anyone at your RTO deciding to adopt them?
Do your trainers and assessors use RTO approved tools only? Or are they free to use public tools in their out-of-class activities?
If a student's evidence looks AI-generated, does your process go straight to a finding, or does someone talk to the student first?
Does your enrolment paperwork tell students how AI may be involved in handling their information?
Does your RTO use a consultant who responds to the regulator on your behalf? And do you oversee their responses?
To help you look deeper into your processes, we have developed (with AI assistance) a review tool that you can use to begin to identify where AI is used in your RTO and how you may be able to improve documentation, rulings and oversight, and support your RTO to adopt the 5 principles for the responsible use of AI in VET.
Our QMS updates that include reference to AI adoption and use will be available on our website soon.
We will also be launching AI use informative support courses which, once ready, will be available here:
It is of utmost importance to me as the Founder and Principal of Coast Wide Training Solutions, that AI is used as a tool, not a solution. That anything we produce looks, feels and sounds like a human. That it is finalised by a human. We fully understand the benefits of AI tools and know that they can be used to get us to 70% of where we wanted (which really, is exactly where we would have gotten to, but it would have taken us 4 times longer) and then we take it the rest of the way.
We always include human oversight in our AI use.
Disclaimer: Claude was used in the development and refinement of this blog, with me, a human having the final say.
Jodie Almond
Principal VET Compliance Consultant
Comments