No Single Rulebook: The 101 on AI in Australian Classrooms

by | Sep 1, 2026 | Education in the Age of AI

Skip to main content
< All Topics
Updated
ByThe Good

Most people's picture of "AI in schools" is still a single image: a kid asking ChatGPT to write their essay. That's real, but it's a small part of what's actually happening. By 2026, generative AI has moved from a novelty a handful of students were sneaking past teachers to something built into national curriculum programs, state assessment rules and the daily workflow of a majority of the profession — and, at the same time, into a live fight over whether some of that access should be pulled back. Before the debate about what AI is doing to education can go anywhere useful, it helps to know what's actually already in place, who decided it, and how fast the ground is still moving underneath it.

From blanket bans to a national framework — in about eighteen months

The starting point matters, because it wasn't policy, it was panic. When ChatGPT launched publicly in late 2022, several state education departments — NSW, Victoria and Queensland among them — responded within months by simply blocking generative AI tools on public school Wi-Fi networks. It was a blunt instrument, and it didn't last: officials came to see the block-it approach as roughly as workable as banning calculators in the 1980s, and the states unwound it in favour of something more structured. By 2026, every state and territory education department has aligned to a shared national approach, and AI tools are now conditionally permitted in classrooms rather than blocked outright, provided use follows academic-integrity and ethical-use rules set at the state level.

There is no AI law for education — there's a stack of overlapping ones

What replaced the ban is not a single law. Unlike workplace safety or medical registration, no piece of Commonwealth legislation governs how AI is used in an Australian classroom or lecture theatre. Several layers apply at once, and no one of them owns the whole problem. The Department of Education runs a national framework endorsed by state and territory education ministers, not legislated by parliament. Universities answer to the Tertiary Education Quality and Standards Agency (TEQSA) on assessment integrity. State and territory education departments set their own binding rules for the schools they run. Privacy law and the eSafety Commissioner sit on top of all of it. That layered, nobody-fully-in-charge structure is a theme this series will return to directly in the governance article — but it's worth flagging here first, because it explains why "the rules" can mean something different depending on which state, sector or institution you're standing in.

The national framework: six principles, no teeth

The centrepiece is the Australian Framework for Generative AI in Schools, built around six principles — teaching and learning, human and social wellbeing, transparency, fairness, accountability, and privacy, security and safety. Education ministers endorsed a review of the framework in mid-2025 that found it still fit for purpose, and working groups are now developing national expectations for the edtech AI tools vendors sell into schools. But the framework itself isn't binding. It's guidance that state systems and individual schools are expected to translate into their own policies — and they've done so unevenly, both in tone and in substance.

NSW's approach runs through the NSW Education Standards Authority (NESA), which built a formal AI policy in 2024 stating that unapproved AI use in assignments is a breach of academic integrity; students must complete a compulsory module called All My Own Work before they're even eligible to take HSC courses. Queensland ties its guidance to the Queensland Curriculum and Assessment Authority (QCAA), which requires all senior students to complete a dedicated academic-integrity course as part of protecting the Queensland Certificate of Education. Western Australia's curriculum authority, the SCSA, has written AI-specific guidance directly into how it assesses the WACE. South Australia has taken a genuinely different route: rather than leading with restriction, the SA government partnered with Microsoft to build its own ChatGPT-style classroom tool, called EdChat, putting a government-sanctioned AI assistant directly into public education instead of managing student access to commercial tools. Four states, four distinct starting philosophies, one shared framework sitting above all of them with no power to force consistency.

Teachers are already using it — more than you'd think

The OECD's TALIS 2024 survey found 66% of Australian lower-secondary teachers had already used AI in their work, the fourth-highest proportion of the 38 OECD member countries surveyed. That's not a fringe behaviour; it's closer to the norm. But the same data shows the profession is genuinely split on what it means: 87% of those teacher-users flagged AI enabling students to misrepresent their own work as their primary concern, while 75% of non-users said they simply didn't have the skills or knowledge to start. Enthusiasm and anxiety are running through the same workforce at the same time, often in the same staffroom — and, increasingly, industry is stepping into that gap: Microsoft's Elevate for Educators program launched in April 2026, offering free training and practical resources to help Australian teachers and school leaders adopt AI safely, precisely because time constraints and a lack of mandated professional development were leaving many teachers to work it out on their own.

The national literacy push, and its gaps

Term 1 2026 saw the rollout of Australia's first locally developed AI literacy curriculum, built by the not-for-profit Code for Schools with Amazon Web Services, delivered through a centralised learning platform to public, independent and Catholic schools, targeting a million students within three years. It covers ethical foundations, practical AI-tool experience, real-world applications and critical-thinking scenarios, with resources for teachers, parents and carers. Code for Schools CEO Megan Woff framed it as a response to teachers facing "time constraints and a lack of mandated professional development" in building AI confidence — a genuine, funded, national commitment. But independent analysis has flagged what's missing: no specified minimum teacher training before rollout, no accountability mechanism to check implementation quality is consistent across schools, and no explicit funding to close device or internet gaps in schools that don't already have them. A curriculum is not the same thing as the infrastructure and training needed to teach it well — a gap the equity article in this series digs into properly.

Security is a second, quieter front

Curriculum and assessment aren't the only place AI policy is moving. Following a Commonwealth directive barring federal public servants from installing DeepSeek — the China-developed AI app — over national security concerns, NSW, South Australia, Queensland, Western Australia, the ACT and the Northern Territory all moved to ban it on government devices, which in practice includes school and department ICT systems. It's a reminder that "AI in education" isn't only a pedagogy question; it's also a live cybersecurity and data-sovereignty one, running on its own track largely separate from the curriculum debate.

Sitting underneath all of it: a system already under strain

None of this policy is landing on a stable, well-resourced system. The ABS's 2025 Schools report — released March 2026 — recorded 4,160,918 students across 9,673 Australian schools, the highest enrolment on record. AI is arriving into classrooms that are simultaneously bigger, more complex and short-staffed — which is exactly why some schools are embracing it as relief and others are approaching it as one more unfunded, unproven mandate. Steiner Education Australia made that second calculation explicit in August 2026, publicly declining to build its own AI learning tool and citing risks to students' critical thinking and independent problem-solving as the reason.

The question worth sitting with

Australia has, in effect, chosen to regulate a genuinely disruptive technology in education the same way it regulates almost everything else about schooling — nationally coordinated in principle, state-run and unevenly resourced in practice. NSW is now moving toward restriction, South Australia toward a government-built alternative and a full royal commission, and Queensland toward a compliance-course model — three different bets happening in the same country, at the same time, under the same six-principle national umbrella. Whether that's the right model for something moving this fast, or whether education needs a more centralised, better-funded response than the usual federal-state compromise, is the question the rest of tonight's material keeps circling back to.

Sources

Table of Contents