AU university AI submissions run at half the US rate

Digital Attitudes, 16 July 2026

Three stories that pull in the same direction. Australian universities finally have hard numbers on how much of a student assignment is AI, an American classroom comms tool goes looking for the principal's desk, and a small US state shows one way to move on AI without waiting for a national policy.

Turnitin puts Australian AI heavy uni submissions at 10.2 per cent, half the US rate

Turnitin's Q2 Learning Integrity Insights Report, released 7 July, analyses 3.1 million Australian tertiary submissions from October 2025 to April 2026 and finds 53.6 per cent used some form of AI, with 10.2 per cent scoring above 80 per cent AI written. That is roughly half the US rate (19.4 per cent, drawn from 33 million submissions) and slightly above the UK (9.8 per cent, from 2.1 million). The same report has 48 per cent of Turnitin webinar participants naming teaching and learning leaders as the primary drivers of AI adoption on campus, well ahead of IT and technology leadership at 17 per cent and cross functional committees at 16 per cent (Turnitin press release, 7 July). Writing in The Conversation on 13 July, CQUniversity's Meena Jha and UTS's Amara Atif read the numbers as a case for assessment redesign rather than sharper detection, recommending unit level tasks where students document how they used AI, evaluate its outputs and reflect on what they accepted or rejected, and calling for education specific AI systems in place of generic ChatGPT (The Conversation, Jha and Atif, 13 July).

Why this matters: For every DVC Education, TEQSA officer and Academic Board member who has spent the year fielding "our students are cheating with AI" emails, the Turnitin number is the first hard cross jurisdictional benchmark that the Australian rate sits at roughly half the American one, on a sample large enough to be worth quoting in a senate paper. That cuts against the moral panic framing in a way "our internal detection rates are up" never did, and it will land inside the Term 2 university leadership meetings that decide whether next year's academic integrity spend goes to more detection tooling or to redesigned unit outlines. Jha and Atif's framing is the one to steal, because the practical implication (a portfolio task per unit that logs student AI use as it happens) matches how Anthropic's Australia usage data already reads and matches what Sydney's Cogniti team have been shipping since 2024. The Turnitin governance number is worth putting on the front page of any AI paper the executive is drafting for council: teaching and learning leaders own AI adoption three to one over IT, which is either an argument for moving your AI budget out of the CIO's cost centre or an argument for putting your CIO on the L&T committee, but it is no longer an argument for leaving the decision with procurement.

ClassDojo goes schoolwide and moves into the principal's comms stack

ClassDojo, the classroom messaging tool used in 95 per cent of US schools and quietly ubiquitous in Australian primary schools, announced on 13 July that its 2026-27 release adds multichannel announcements (email, text, voice, mobile app), Canva powered newsletters, one click social posting, schoolwide engagement dashboards and, for the first time, tools for middle and high school staff to message students directly. Auto translation covers 190 languages and the leader features stay free. General Manager Jeff Buening pitched the release as fewer tools that work better together, and cited internal figures that 80 per cent of families read a ClassDojo message the day it lands, with schoolwide implementations seeing 25 per cent more active families and 20 per cent more family initiated messages (ClassDojo release, 13 July).

Why this matters: For an Australian primary principal, ClassDojo has been the family channel that actually works while the SIS bolt on (Compass Inklass, Sentral parent portal, School Bytes messaging) has been the one on the vendor contract, and the new schoolwide product line means the day one comms conversation now starts with a free tool that already has 80 per cent open rates and finishes with a paid one that does not. The secondary school angle is the bigger one, because it is the first credible pitch to move ClassDojo out of the primary sanctum and into a market where Compass, Sentral and SEQTA have been the default parent surface for a decade, and every one of those platforms has a communications module they have been trying to charge for. Expect the family engagement number to appear in ClassDojo's Australian school pitch inside a fortnight; if you run comms at a primary or a K-12 independent, the honest question for the next SIS renewal is whether your family messaging module is worth what you are paying if a free tool with translation into 190 languages is now doing the newsletter, the Canva design and the engagement dashboard on one login.

Maine sidesteps a national style AI policy for a six step district framework

Maine's Department of Education presented its schools AI approach at the Education Commission of the States National Forum on 9 July, and the interesting choice is what it did not do. Coordinator of Learning Through Technology Emma-Marie Banks and Chief Teaching and Learning Officer Beth Lambert told the forum that Maine deliberately skipped the traditional identify, draft, review, approve, implement policy cycle because "by the time you finish that cycle, the technology had already changed" (Lambert), and instead built a six step framework grounded in the state's existing computer science structure, layered with an updatable toolkit, 1:1 coaching, district leadership meetings and shorter professional development modules that revise as educators feed back (K-12 Dive, 14 July). Banks framed computer science as "the foundation of which AI is built" to lower teacher anxiety before generative tools get introduced; Lambert flagged the state's philosophy did not change with AI, "student success was still the goal", and that AI is being handled as a cross cutting issue rather than a standalone ed tech question.

Why this matters: For every Australian state education secretary reading yesterday's Clare communique, Maine is the live case study of what it looks like to move on AI without either a national framework document or an anything goes vacuum, and it is worth putting alongside the NSWEduChat rollout, South Australia's EdChat and the ACT's Gemini/Copilot pilots as a comparator when the department drafts its next AI direction. The federal Office of AI in PM&C is going to write the ceiling framework, and every state department is now deciding how much of its own governance stack it hands upward and how much it keeps close, and the Maine model (foundation subject first, toolkit second, PD third, formal policy last) is a practical answer that has stayed durable for three years without a single monolithic document to defend at estimates. The bit worth stealing for an AU state is the 1:1 coaching and district leadership meetings, because those are the parts of the AU state model that Corella (Queensland), the TAFE NSW Gen AI policy review and Victoria's post breach recovery all lack, and they are cheaper to fund than another portal.

Term 3 opens Monday. Numbers, comms tools and one small state's playbook. More at digitalattitudes.com.au.