For Australian teachers and school leaders | Updated 26 October 2026. This practical guide explains the six principles of Australia's national generative AI framework and offers classroom applications, not a substitute for your state, territory, school-sector or school-specific rules.
A Year 8 student opens an AI chatbot to check a paragraph about the causes of Federation. Within seconds, the screen displays a polished answer, complete with confident-sounding explanations. The student is pleased. The teacher is less certain. Did the learner clarify an idea, or simply replace the thinking that the lesson was designed to develop?
That question sits at the heart of Australia's approach to artificial intelligence in schools. The real challenge is not choosing between a blanket ban and unrestricted enthusiasm. It is deciding which learning should remain unmistakably human, where technology might genuinely help, and who takes responsibility when it does not.
The Australian Framework for Generative Artificial Intelligence in Schools was approved by Education Ministers in October 2023 and implemented across school systems from Term 1, 2024. Ministers endorsed a review in June 2025 that found the framework remained useful. In October 2026, the Australian Education Research Organisation (AERO) added an important reality check: robust evidence about the effects of school AI use on learning remains limited. Its advice asks teachers to prioritise established approaches to effective learning, watch for risks and establish explicit expectations.
For a busy teacher, that means the framework is neither a shopping list of AI applications nor a requirement to put chatbots into every lesson. It is a set of decisions about learning, relationships, honesty, inclusion, judgment and safety.
What are Australia's six AI-in-schools principles?
The national framework is organised around six linked principles, supported by 25 guiding statements. The names are official; the classroom practices below are Learning Coach's practical interpretation, not additional government rules.
| Official principle | The question a teacher should ask |
|---|---|
| Teaching and Learning | Does this use develop understanding or substitute for it? |
| Human and Social Wellbeing | Does it protect relationships, confidence and belonging? |
| Transparency | Can students and families understand when and how AI is used? |
| Fairness | Who might be excluded, misrepresented or disadvantaged? |
| Accountability | Which human checks, decides and corrects? |
| Privacy, Security and Safety | Is the tool approved, appropriate and safe for the data and age group? |
The Australian Curriculum's AI connection also distinguishes learning about AI from using AI to help learn something else. Students need opportunities to understand how systems work, recognise their limitations and practise responsible use. But learning to prompt is not the same as learning history, science, writing or mathematics.
Principle 1: Teaching and Learning — keep the learner doing the thinking
The first principle is simple to state and demanding to apply: generative AI should enhance teaching and learning. An impressive output is not evidence that a student has developed knowledge or a skill.
Consider a Year 9 mathematics lesson on linear relationships. A learner asks AI to solve six equations and receives six correct worked solutions. If the teacher accepts the output, the apparent success conceals an unanswered question: could the student solve a new equation alone, explain why each operation is valid and notice an incorrect step?
A stronger sequence starts with the learner attempting a problem independently. The teacher observes where understanding breaks down. An approved AI tool, when permitted, might then supply a different worked example, a question or a hint rather than the answer. The learner returns to the original task and solves a comparable problem without support.
This is the distinction between assistance that develops capability and assistance that conceals its absence.
Put it into practice
Start lesson planning with the capability, not the tool: "Students will be able to justify an inference from a historical source", rather than "Students will make an AI presentation". Identify what learners must retrieve, discuss, calculate, write or explain for themselves. Only then ask whether an approved AI application adds value.
A useful teacher question is: "If I remove the chatbot during the final five minutes, what can each student still demonstrate?" That final check might be a short handwritten explanation, a new problem, a verbal justification or a comparison of two sources. It should test the lesson's objective, not simply whether students can repeat the AI's words.
AI can sometimes assist a teacher in drafting differentiated examples or generating ideas for formative questions, but the teacher must check the accuracy, difficulty, cultural context and curriculum alignment of every resource. The time saved is only valuable if the resulting lesson improves opportunities to learn.
Evidence to collect: a first independent attempt, a record of feedback or support used, and a final independent explanation. These three points tell you far more than the polished submission alone.
Principle 2: Human and Social Wellbeing — protect relationships and belonging
A classroom is a social environment before it is a technological one. Students learn with peers, ask awkward questions, make mistakes, recover confidence and discover that adults are interested in their thinking. An AI assistant can simulate conversation; it cannot replace the teacher's duty of care or the relationships through which students experience school.
Imagine a quiet student who is reluctant to share a draft with classmates. An AI system may offer a low-pressure opportunity to rehearse ideas, if the school's rules permit it. But if every opportunity for rehearsal migrates onto a screen, the student might get fewer chances to practise explaining ideas to real people. The goal should be to use any private preparation as a bridge to human participation.
The same care applies when a student asks a chatbot about bullying, anxiety, safety or a difficult family situation. A general-purpose AI tool is not a school counsellor, child-protection contact or emergency response service. Teachers need established safeguarding and referral procedures, not improvised chatbot conversations.
Put it into practice
In collaborative tasks, reserve clear roles for human discussion: one student questions an explanation, another offers a counterexample and another summarises the team's reasoning. If an approved AI tool is used, its contribution is examined as a resource, not treated as an additional classmate or authority.
Ask students how the activity affected their confidence and willingness to contribute. Check who is participating more and who has become less visible. Preserve screen-free moments for reflection, relationships and sustained attention. For younger pupils, use extra caution and follow age-specific departmental guidance and product restrictions.
A wellbeing-aware AI routine ends with a human exchange: "What do you think now? What would you ask someone else? What still confuses you?"
Evidence to collect: student participation, help-seeking, peer discussion and the teacher's observations—not a promise that a particular tool will improve wellbeing.
Principle 3: Transparency — make AI assistance visible and understandable
Transparency is not achieved by announcing that "our school uses AI". A student should be able to tell when AI is involved, what it was used for, which limitations matter, and what they are expected to disclose in their own work.
Suppose a Year 10 student submits a well-structured persuasive essay. The teacher allows AI-generated vocabulary suggestions but not AI-written arguments. Without a shared disclosure routine, the student may genuinely be unsure whether the submission follows the rules.
Replace vague instructions such as "use AI responsibly" with task-level expectations: what is allowed before writing, what is allowed during drafting, what must remain the student's work and what must be acknowledged.
Put it into practice
Use a brief, age-appropriate disclosure statement with three prompts: "What tool did you use, if any? What did it help you do? Which claims or changes did you personally check?" For younger learners, a verbal explanation may be more suitable than a written log.
Show students how generative AI can produce fluent yet inaccurate answers. Put a teacher-prepared, non-sensitive example on the board, verify it using reliable sources, and discuss why confidence in the wording is not proof. This links directly to the Australian Curriculum's emphasis on understanding how AI works and its limitations.
When teachers use an approved system to prepare examples, worksheets or feedback prompts, they should also follow their employer's disclosure and quality-assurance requirements. Transparency is a two-way expectation.
Evidence to collect: understandable assignment directions, proportionate disclosures and students' ability to explain their own work.
Principle 4: Fairness — check access, bias and whose knowledge is represented
An AI activity can look equal while producing unequal learning opportunities. Some students have reliable devices and home internet; others do not. Some students can independently challenge an inaccurate answer; others cannot yet distinguish a plausible explanation from a correct one. Some cultural and linguistic contexts may be misrepresented by the system's output.
Fairness therefore means more than offering the same chatbot to everyone. It requires considering accessibility, alternative routes to the learning goal, bias in generated material and the consequences of mistakes.
Take a Year 7 geography activity about Country and Place. If a chatbot produces a generalised account of Aboriginal and Torres Strait Islander peoples' knowledge, the teacher cannot treat that text as an authoritative substitute for properly selected local sources or appropriate community perspectives. AI output may be incomplete, inaccurate or culturally insensitive.
Put it into practice
Audit the activity from the perspective of a student without a personal device, a student using assistive technology, a multilingual learner and a student who needs additional scaffolding. Ask whether each can meet the same objective through an approved and accessible route, including a non-AI alternative.
Check AI-generated examples for stereotypes and unsupported claims. Use authoritative teaching materials, appropriate First Nations resources and relevant local consultation rather than asking a model to speak for a community. Build in opportunities for students to challenge representations and cite reliable evidence.
Do not confuse identical treatment with equitable participation. Some students need more explicit teaching before an AI-supported activity becomes useful; others may need fewer digital prompts and more independent challenge.
Evidence to collect: access and accommodation decisions, checked resources, learner feedback and signs that every student has a real opportunity to demonstrate understanding.
Principle 5: Accountability — human teachers own the decisions
Accountability asks a question that excitement about automation can obscure: when an AI system is wrong, who notices, who can challenge the result, and who corrects it?
A teacher may use an approved system to brainstorm a rubric, draft a quiz or suggest possible misconceptions. That does not make the system accountable for the quality of assessment or the fairness of decisions affecting students.
An AI-generated comment such as "This learner shows weak reasoning" can be misleading if it is based on limited work, incorrect criteria or a poor understanding of the student's learning history. Teachers and school leaders remain responsible for professional judgments within their roles and applicable policies.
Put it into practice
Before using an AI-assisted assessment resource, ask three questions. Is the information accurate? Does the judgment reflect the actual task and curriculum standard? Can I explain the decision to the student and family without appealing to "the AI said so"?
Keep meaningful human review where it matters most: grades, behaviour or welfare judgments, referral recommendations, personal learning plans and high-stakes communications. Provide ways to contest errors. Prefer tools and processes whose operation, provenance and limitations staff can explain.
School leaders should also distinguish approval of a product from approval of every conceivable use of that product. A department-approved assistant for lesson planning may not be appropriate for individualised student assessment or confidential pastoral records.
Evidence to collect: documented review decisions, corrections made by teachers and clear processes for raising concerns.
Principle 6: Privacy, Security and Safety — treat student information as protected
Every prompt sends information somewhere. That fact matters greatly when dealing with minors, assessment records, personal circumstances, photos, disability-related adjustments or information about families.
The national framework places privacy, security and safety alongside pedagogy, not after it. Schools must follow the applicable privacy, procurement, cyber safety, safeguarding and departmental requirements. Availability on a public website does not mean an AI service is approved for student use.
The NSW Department of Education's published guidance, for example, recommends its department-developed NSWEduChat and cautions staff about entering student work or personal, sensitive and confidential data into tools that have not been approved for that purpose. Other jurisdictions and sectors have their own procedures.
Put it into practice
Before any classroom pilot, confirm product approval, permitted ages, login arrangements, data-handling requirements and the exact use case. Never paste identifiable student material into a public chatbot on the assumption that a prompt is private. Where policy permits use of a tool, minimise the data shared and remove identifying details unless an explicitly authorised workflow requires otherwise.
For a lesson on persuasive writing, use a fabricated or de-identified sample provided by the teacher instead of uploading a student's actual confidential work. Do not ask students to create accounts for tools that the school has not approved.
Teach students a simple boundary: a helpful-looking website is not automatically a safe place for your name, image, schoolwork or personal story.
Evidence to collect: approved-tool checks, documented data practices, age-appropriate instructions and an incident escalation process.
What does a responsible AI lesson actually look like?
The six principles become useful when they change the flow of a lesson. Here is one practical option for a 45-minute Year 8 science lesson investigating the reliability of claims about energy transfer. It is a suggested teaching routine, not an official government lesson template.
| Phase | Teacher and student action | What it protects |
|---|---|---|
| First 8 minutes | Students explain an energy-transfer example without AI | Baseline understanding |
| Next 8 minutes | Teacher models how an approved tool might answer the same question | Transparency and AI literacy |
| Next 10 minutes | Pairs compare the generated answer with trusted class resources and identify errors or omissions | Accuracy and critical thinking |
| Next 10 minutes | Students revise their own explanation, keeping a simple record of what changed and why | Ownership and accountability |
| Final 9 minutes | Students explain a new case without AI and answer a teacher follow-up question | Independent transfer |
The strength of this design is that students do not merely evaluate a chatbot. They must use evidence and then perform the target thinking again. If the approved tool adds no meaningful benefit, the teacher can run the same comparison with a prepared example on paper.
The method also creates a record of where support helped. A learner who cannot explain the new example after using AI needs additional teaching, feedback or practice—not simply a more sophisticated prompt.
A four-week implementation plan for Australian teaching teams
Individual teachers need clear classroom routines, but responsible use cannot depend entirely on each teacher inventing their own rules. Departments, systems and school leaders set the boundaries within which professional judgment operates. The following plan fits a team preparing a small pilot within its existing policy.
Week 1 — Define the learning problem
Choose one curriculum objective and one pain point: perhaps students struggle to explain why a mathematics method works, or the teacher needs more varied examples for a mixed-attainment group. Write the learning outcome before discussing products. Check current departmental policy, age requirements and approved tools.
Ask: "What will students know or be able to do that they could not do before?" Avoid using prompt count, login frequency or enthusiasm as the success measure.
Week 2 — Build the safeguards
Prepare a concise permitted-use statement for students and families, a non-digital alternative, the teacher verification procedure and a straightforward disclosure routine. Agree on what information will never be entered into a tool. Make sure colleagues understand how to report errors, inappropriate outputs or safeguarding concerns.
Check the activity for accessibility and cultural responsiveness. Remember that a tool being popular among teachers is not equivalent to it being evaluated and approved by the school system.
Week 3 — Run one bounded lesson
Pilot one lesson, not a wholesale change to homework or assessment. Collect students' unaided first responses and unaided final responses. Note whether AI prompts helped produce explanation, questioning and revision, or simply made work appear more polished.
Ask learners to explain what they trusted, what they challenged and what they changed. Include students who chose the non-AI route in the evaluation.
Week 4 — Review before expanding
Compare the evidence with the original objective. Was understanding clearer? Could students explain a fresh example without assistance? Did time spent checking outputs cancel time saved creating them? Were any learners disadvantaged? Were there privacy or wellbeing concerns?
Record an explicit decision: continue, modify, pause or stop this use case. Keep the review short enough for colleagues to repeat. A responsible pilot can conclude that a conventional teaching approach is preferable.
This measured approach is consistent with AERO's October 2026 advice, which stresses evidence-based teaching, clear expectations, assessment design and culturally responsive practice in a field where robust evidence of learning impact remains limited.
AI assessment rules: why task design matters more than detection
Teachers are sometimes offered a false choice: permit all AI assistance or spend evenings trying to catch students using it. The better question is how an assessment can make the intended learning observable.
If an assignment is designed to measure reasoning, the final answer alone is insufficient evidence. Consider incorporating in-class drafting, brief oral explanations, process notes or a new transfer task. Use methods proportionate to the subject, year level and assessment requirements.
The changing policy environment is a reminder that local rules matter. On 1 September 2026, New South Wales announced restrictions on the number and weighting of take-home HSC assessments as part of its response to AI-related integrity concerns. That state-specific change does not rewrite the national six principles or automatically apply to every Australian school task. Teachers should check the current requirements from their authority rather than generalising from a headline.
For ordinary classroom work, a useful rule is: AI may support a process only when the teacher can still obtain trustworthy evidence of the student's own learning.
The teacher as coach: helping students become less dependent on assistance
Technology makes it possible to obtain a polished answer immediately. Learning still requires attention, effort, feedback and the ability to use knowledge without constant support. This is where a teacher's coaching stance becomes particularly relevant.
A teacher acting as a learning coach does not abandon instruction. They combine expert teaching with careful observation, questions that reveal thinking, feedback that helps the student act, and a gradual transfer of responsibility. Instead of repeatedly correcting an AI-assisted paragraph, the teacher might ask, "Which sentence can you defend with evidence? How would you explain it without looking? What will you check next time?"
The important movement is from dependence towards conscious, repeatable strategies. The learner notices what was difficult, uses an appropriate resource and becomes increasingly capable of doing the work independently.
Learning Coach explores this approach through its training and specialisation pathway, including work relevant to teachers who want to strengthen students' autonomy and learning methods. Readers looking for a broader introduction to the learning-coaching philosophy can also explore Become a Learning Coach. These are supplementary professional-development resources, not Australian government endorsements or claims that a particular training route is required for framework compliance.
Questions Australian teachers frequently ask
Is generative AI compulsory under the Australian framework?
No. The framework guides responsible and ethical use; it is not a mandate to introduce AI into every classroom activity. A teacher can decide that a task is better taught without AI, subject to relevant school policy and curriculum obligations.
Does a national framework replace state or territory rules?
No. State, territory and non-government school systems implement their own guidance and arrangements. Check your employer's rules on approved tools, student accounts, data protection, parent communication, assessment and safeguarding before introducing an activity.
Can students use AI to draft essays or complete homework?
There is no single permission that fits every task and jurisdiction. The teacher or school must make acceptable uses explicit. If independent writing is being assessed, AI-generated prose may undermine the evidence the task is meant to capture. Students need clear directions about permitted support and any disclosure required.
How can a teacher tell whether AI improved learning?
Compare independent student performance before and after support and check whether students can explain, apply and transfer ideas without the tool. A faster submission or a nicer paragraph does not by itself show durable learning.
The real test of an AI-ready classroom
Australia's six principles offer a sensible sequence for classroom decisions: start with the learning goal, protect the people involved, make the technology visible, ensure fair access, keep humans accountable and safeguard information.
A school does not become AI-ready by adopting the most powerful chatbot. It becomes ready when teachers can justify why a particular use belongs in a lesson, students know what thinking remains theirs, and leaders can recognise when a technology should be limited or withdrawn.
The question worth taking into the next staff meeting is not "What else can AI do?" It is "What will our students be able to do for themselves?"
If your goal is to develop that independence more intentionally, explore the Learning Coach training pathway for educators and consider how an observation-led, coaching-informed approach could complement your current teaching practice.
Resources and references
Australian Government Department of Education — Australian Framework for Generative AI in Schools. Official national framework and 2025-endorsed review.
Australian Government Department of Education — 2024 Framework Review, announced June 2025. Official review summary.
Australian Curriculum, Assessment and Reporting Authority — Artificial intelligence curriculum connection. Curriculum guidance and the six principles.
Australian Education Research Organisation — AI in schools: Considerations for school leaders and teachers. Evidence explainer, 14 October 2026.
NSW Department of Education — Guidelines regarding the use of generative AI. State-system implementation example.
NSW Government — New rules for schools on AI in take-home HSC assessments. Ministerial release, 1 September 2026.
This article is educational guidance, not legal, regulatory, procurement or safeguarding advice. Follow the current rules of the applicable Australian jurisdiction, school sector and school.
