ReadyTech names AI integrity gap as ACER ships Foundation reading test

Digital Attitudes, 17 August 2026

Monday of Term 3 Week 6, and three stories from the past week land squarely on the Term 4 planning table. An Australian ed-tech vendor has audited its own customers and put a number on the AI integrity doubt, ACER has shipped a diagnostic that pushes reading assessment down to Foundation, and Singapore's largest university has done what no Australian sandstone has yet done, made a generative AI module compulsory for every incoming undergraduate.

ReadyTech's AI benchmark puts a number on the doubt Australian ed-tech has been dancing around

The ReadyTech AI and Innovation Benchmark 2026, published on 14 August, surveyed 96 Australian education organisations (42.7 percent higher-ed, the rest schools and Registered Training Organisations) and reports 43.8 percent expect AI-related academic integrity issues to worsen over the next two years, with 73 percent already treating the problem as significant, very significant or critical (The Educator on the ReadyTech AI and Innovation Benchmark). Only 34.4 percent have a documented AI use policy and 30.2 percent have a responsible AI framework, while 42.7 percent are deploying AI detection tools despite the sector's growing verdict (Curtin in January, UQ in mid 2025) that detectors are unreliable. ReadyTech CEO Marc Washbourne's line in the release is telling: "Too many organisations still lack clear productivity baselines and don't consistently measure the value AI delivers."

Why this matters: this is a customer facing Australian ed-tech vendor auditing its own market and publishing the answer nobody wanted on the record. For a school leader or a university CIO writing a Term 4 board paper, this is the Australian data point to cite alongside Furze's mood shift piece (Saturday's brief) and the ReadyTech CEO's admission that vendors do not know what value AI is delivering. Two concrete actions. First, the roughly two thirds of organisations without a documented AI use policy have run out of runway; NESA's HSC review (Tuesday's brief), the National Cabinet AI standards process (Monday's brief) and the SA Royal Commission (Friday's brief) will each reference organisation level policies in whatever they produce. Second, the 42.7 percent still deploying AI detection are stacking risk against a benchmark that names the practice; the ReadyTech paper is now the polite way to raise that around the table without picking a fight with the head of learning.

ACER launches PAT Early Literacy for Foundation and Year 1, and the timing is not an accident

The Australian Council for Educational Research announced PAT Early Literacy on 11 August 2026, a digital assessment for Foundation and Year 1 that measures phonics and related skills, listening comprehension and reading comprehension separately, with reading comprehension results tied to the existing PAT Reading scale so schools can track growth on a continuous measure from Foundation onward (The Educator on the ACER PAT Early Literacy launch). Senior assistant director of assessment development Sandra Knowles framed the design in the release: "Reading is more than phonics alone. By assessing phonics and related skills, listening comprehension and reading comprehension separately, teachers can better understand what is driving a student's literacy development and make more informed decisions about next steps." The launch lands six days after ACARA's NAPLAN 2026 release reported reading and writing performance declining across most year levels even as numeracy improved.

Why this matters: the operational value is that the Australian sector now has a nationally scoped early years literacy diagnostic tied to the same scale that Year 3 to 10 teachers already read from. For a K to 12 head of teaching and learning, that means Foundation and Year 1 data can go into the same longitudinal chart as the NAPLAN and PAT Reading data the school already reports on, rather than sitting in a separate spreadsheet the deputy principal never opens. The Term 4 question is whether the school buys in for a 2027 Foundation baseline, and the answer for most is probably yes, because the alternative (waiting until Year 3 NAPLAN to know if a cohort is behind) is exactly the pattern the widening parental education gap (Friday's brief) argues against. For ACER, this is also a defensive move against MultiLit's InitiaLit assessment and against the state supplied phonics screening checks that Victoria and NSW have already rolled out; the differentiator is the vertical scaling to PAT Reading, and that is the sales pitch that a data literate deputy principal will actually respond to.

NUS makes a generative AI module compulsory for every freshman while Australia's sandstones are still writing policy

The National University of Singapore announced on 11 August 2026 that all first year undergraduates from Academic Year 2026/27 will complete THE1008 "Applied Generative AI: From Prompting to Evaluation" as a compulsory module, and that all NUS students, faculty and staff will have ChatGPT Edu access from 31 August under a strategic collaboration with OpenAI signed on 6 August (NUS on the OpenAI strategic collaboration, Mothership on the NUS ChatGPT rollout). Deputy President (Academic Affairs) Aaron Thean's release line is that the collaboration is "a pivotal step in our human centric AI strategy" and OpenAI's Oliver Jay confirmed the earlier School of Computing deployment (March 2026) has shown "higher engagement and adoption rates than similarly sized universities." Data does not leave the university managed workspace and is not used to train OpenAI models.

Why this matters: this is the direct comparison point for every Australian university procurement team currently weighing what to require of first year students in 2027. UNSW has the 10,000 licence ChatGPT Edu staff deployment (September 2025), La Trobe is stepping through 5,000 licences in 2026 to 40,000 by 2027 and is building the AI MBA, Monash is on Microsoft Copilot for students, and USYD's Green Paper (Wednesday's brief) argues the assessment redesign task rather than a compulsory module. NUS has picked a fourth option, mandatory foundational AI literacy for every incoming student paired with an enterprise grade ChatGPT deployment on the first day of semester. For an Australian pro vice chancellor teaching and learning, the pointed question is whether the local strategy stops at licences and policy or extends to a compulsory first year subject, because Bond's Bachelor of Enterprise AI (Saturday's brief) has already argued undergraduate AI capability belongs inside the degree, not the orientation deck. If the answer is a compulsory subject, the design work needs to start now to be in the 2027 handbook.

Three moves, one week, and the pattern is that the Australian sector is being audited by its own vendor, out shipped in early years assessment by a national research council responding to a NAPLAN gap that widened this month, and out mandated at the undergraduate level by an Asian peer. More on the operational landscape at digitalattitudes.com.au.