SA calls AI royal commission with education in scope, NAPLAN gap widens

Digital Attitudes, 14 August 2026

Friday of Term 3 Week 5, and Adelaide has just made itself the first Australian jurisdiction to put AI in schools inside a royal commission, ACARA's 2026 NAPLAN release has a fine print worse than the headline, an Australian relief teaching platform is exporting itself to Dublin, and Blackboard has quietly rewritten the Gradebook grid mid semester.

SA calls a $3m AI royal commission and puts schools inside the terms of reference

Premier Peter Malinauskas announced on Monday 11 August that South Australia will hold the country's first royal commission dedicated to artificial intelligence, with education, health, creative industries, jobs and the public sector all named in the terms of reference (ACS Information Age on the SA AI royal commission, Cyber Daily coverage). The estimated cost is above $3 million, three commissioners will be appointed with mixed expertise in AI, labour economics and business, the commission formally launches on 1 October 2026, and the final report is due to government no later than 1 July 2027. Malinauskas's own framing puts schools front and centre: "in a regulated way, we want to teach kids how to use AI but I don't think we've done as much work as we potentially can or should." The Opposition has already questioned the cost, with concerns filed in InDaily (InDaily on the royal commission cost debate).

Why this matters: SA has just moved AI in education from the framework layer (Australian Framework for Generative AI in Schools, the department's own EdChat guardrails) to the royal commission layer, which is a much heavier instrument and produces sworn evidence, published exhibits and public hearings. For a principal or a director of digital learning in SA public or Catholic, the Term 4 preparation is a written record of what your school actually does with AI today, because it is now reasonable to assume the commission will ask for it. For a vendor selling into SA (Microsoft on EdChat, any external tutoring or marking product used inside a public high school), the same rule applies. For every other state, the practical read is whether the SA commission's evidence base ends up doing the empirical work that AERO and the National Cabinet AI standards process (see Monday's brief) have not been resourced to do. It arrived without a heads up to the PM, which tells you the states are done waiting for a national line.

ACARA's 2026 NAPLAN dashboards show the parental education gap widening at every year level

Andres Molina, senior research fellow at Victoria University's Mitchell Institute, published a fine print reading of ACARA's 5 August NAPLAN release in The Conversation on Monday, and the numbers are worse than the headline reporting suggested (Molina in The Conversation on the widening NAPLAN gap). The Year 3 reading gap between students whose parents completed university and students whose parents did not complete Year 12 grew from 1 year 10 months (2023) to 2 years 2 months (2026). Year 7 reading widened from 4 years 11 months to 5 years 6 months. Year 9 reading widened from 6 years to 6 years 8 months. Year 9 numeracy widened from 6 years 8 months to 7 years 9 months. Molina's summary is direct: "achievement differences emerge early, and then widen in later year levels." The 2026 test covered roughly 1.2 million students across 9,400 campuses.

Why this matters: for a school system running any kind of learning analytics platform, adaptive learning product or catch up tutoring stack, the 2026 NAPLAN dashboards on My School are the raw material for every Term 3 improvement plan and every Term 4 board paper, and Molina's read is that the picture is getting worse for exactly the students most likely to be on the receiving end of the software. The concrete question for a head of teaching and learning this week is not whether the dashboards look bad (they do), it is which specific intervention (a targeted small group program, a specific Compass or Sentral report, a specific adaptive tool) has evidence of closing the parental education gap rather than widening it. If the answer is "we do not know," the Term 4 improvement plan should read "start measuring" rather than "buy more." The Molina piece is short enough to send to a board chair on the weekend and pointed enough to change the conversation.

ClassCover uses its Australian base to bankroll a 3 million euro Irish launch

ClassCover, the Sydney based substitute teacher booking platform founded by Ben Grozier in 2011, announced on Sunday 10 August that it is investing 3 million euro to enter the Irish market, with 20 new roles in sales, marketing, partnerships and customer support, and a target of onboarding 2,000 Irish schools (about half the country) within three years (RTE on ClassCover's Irish launch, Silicon Republic on the ClassCover investment). The Australian footprint the expansion is being funded off is 5,500 schools, 200,000 teachers and SNAs, and 500,000 bookings a year, with a projection to reach 650,000 within 12 months as another 1,300 schools sign on. The Dublin pilot begins later in August. Grozier told RTE the expansion was seeded by the Irish teacher community already using the platform inside Australia.

Why this matters: this is one of a small number of Australian ed-tech companies large enough to internationalise off domestic revenue rather than a fresh venture round, in a sector where ANZ ed-tech funding was down 67 per cent in 2025 (per HolonIQ). For principals and school business managers who have a relief teacher shortage problem (which is most of them), the practical read is that a single Australian vendor now touches 5,500 school offices every morning and is confident enough in the model to run it on a second continent. That has two implications. The first is that ClassCover has become a piece of infrastructure inside the daily attendance and payroll workflow, and the Term 3 question is whether the school's substitute teacher process has a second option in the drawer if the platform ever has a bad morning. The second is that vendor consolidation in Australian K-12 SaaS is happening at the edges (payroll, relief booking, wellbeing) not just at the core LMS, and the 2027 procurement audit should count them properly.

Blackboard rewrites the Gradebook grid mid semester and universities absorb the change

The University of Queensland's central eLearning team posted a notice this month that Blackboard's Gradebook grid view was updated on 7 August, with staff warned to allow time to relearn the layout before Semester 2 assessment cycles land (UQ eLearning on the Blackboard Gradebook update). No data loss, no functional break, but the interaction pattern has changed, and every UQ course coordinator, tutor and casual marker who touches the grid has to adjust while Week 5 assessments are in the queue. UQ is one of the larger ANZ Blackboard Learn deployments, and other Blackboard using tertiary institutions in the region are looking at the same forced UI change over the same window.

Why this matters: the loud LMS story of 2026 has been the assessment redesign conversation and the AI marking layer, and it has quietly overshadowed the ordinary UI churn that vendors ship into a live semester. For a director of digital learning at a Blackboard shop, this is the reminder that the vendor's release calendar does not respect the semester calendar, and the Term 3 operational discipline is to keep a running two week look ahead of vendor product changes across every core platform (Blackboard, Canvas, Moodle, Turnitin, Compass, Sentral) and route the change management task to the right academic development team before the release ships. The other read is procurement. When a vendor changes the primary marking surface in the middle of the semester it uses licensing bought before the change, the contract question at renewal is whether "no material UI change during teaching weeks" is a term that should be written into the schedule.

Four stories, one week, and the direction of travel is unmistakable. AI in Australian education is moving out of the framework debate and into the evidence phase, whether via a royal commission in Adelaide, a NAPLAN dashboard in every state, or a vendor product change that lands whether the semester is ready or not. More on the operational landscape at digitalattitudes.com.au.