AI in Schools
Luckin reframes AI detection, Liu calls for assessment redesign
Digital Attitudes, 15 June 2026
The Sunday read is two reframes of the Australian AI assessment conversation that landed on sector desks this week, one keynote out of EDUtech 2026 and one Deputy Vice-Chancellor in Education Review, plus a third piece on what schools should actually do about the AI tool the detection conversation does not catch.
Luckin tells EDUtech the detection question is the wrong one
Professor Rose Luckin's keynote at EDUtech 2026 in Sydney on Wednesday 3 June asked the room to stop asking "how do we catch students using AI?" and start asking what made AI a rational response for the student in front of you. Teacher Magazine's Rebecca Vukovic and Dominique Russell published their reflections on Friday 5 June, citing Luckin's clean breakdown of what generative AI is and is not good at: pattern at scale, drafting and generating options on the strengths side, knowing truth, caring about persons and first principles reasoning on the weaknesses side. Jason Lodge, on the same panel, named the seventeen year typical lag between research and policy as a problem the sector cannot afford on this one. Teacher Magazine on the EDUtech 2026 reflections, 5 June.
Why this matters: For school principals and university course coordinators sitting on assessment review committees, Luckin's reframe gives the policy room a sharper question than the detect or teach binary Anna Borek raised on Friday, because the question is whether the assessment design produced the AI use, not whether the student should be penalised for using it. For TEQSA reading institutional action plans against its September 2025 Threshold Standards monitoring, the reframe is the line of argument the regulator's next assessment expectation paper will sit on. For ed-tech vendors selling AI detection (Turnitin, Copyleaks, GPTZero, Originality.ai), the procurement question Luckin sharpens is whether the next contract conversation is about detection at all, because the vendors selling assessment redesign tools (Cadmus, Inspera, Möbius) are the ones the policy reframe walks into the room with.
Liu tells Education Review the assessment redesign window is open
Erin Morley at Education Review reported on Tuesday 9 June that Sydney's Deputy Vice-Chancellor (Education) Professor Danny Liu, the architect of the two lane Sydney Assessment Framework and the in-house Cogniti AI platform that hit the AAP wire on Saturday morning, says Australian schools and universities are still using assessment built on assumptions from before ChatGPT. Liu's call: the past three years of technological change should prompt the sector to fundamentally rethink what and how it assesses students, with AI literacy and the competencies the future workforce needs at the centre of the redesign. Education Review on Liu's call for assessment redesign, 9 June.
Why this matters: For university Pro Vice-Chancellors (Education) writing the 2027 assessment plan under TEQSA's regulatory led framework that started this year, Liu's interview reads as the cleanest single sector statement available on what the two lane model actually asks the institution to deliver. For ATEC, picking up the Higher Education Standards Framework on 1 July with Bart Glover at the helm, Liu's argument is the case file the commission will cite when it writes the standards expectation paper on AI literacy as a graduate attribute. For curriculum bodies in the school sector (ACARA, NESA, VCAA, SCSA, QCAA), Liu's framing puts the assessment redesign conversation on the same desk as the PISA 2029 AI literacy assessment ACER landed last week, and the two papers will get read together.
Schools shift the deepfake conversation from banning the app to teaching the student
EducationDaily published a piece on Thursday 4 June arguing Australian schools are moving from blanket bans on generative image and nudify apps to media literacy education that teaches students to spot synthetic media, cross reference claims with verified sources, and treat the eSafety Commissioner's Toolkit for Schools as the operational reference for response. The piece's central observation: bans push the technology into unmonitored home environments and the technology evolves faster than the school's network filter, so the long term answer is the student's discernment, not the school's URL block list. EducationDaily on schools confronting the deepfake frontier, 4 June.
Why this matters: For independent and Catholic school heads writing the 2027 acceptable use policy and ICT strategy, the EducationDaily framing is the sector's emerging line on the question the Luckin and Liu reframes do not answer, which is what to do when the student is producing AI output rather than consuming it for assessment. For eSafety Commissioner Julie Inman Grant's office, which has the Children's Online Privacy Code to register by 10 December and a Toolkit for Schools update on the deepfake response pathway, the EducationDaily piece is the sector's invitation to keep the Commissioner's resources in the procurement conversation alongside the network security stack. For state Departments of Education running URL filter contracts with the major vendors (Cisco Umbrella, Zscaler, Fortinet, Palo Alto), the operational read is that the filter is the floor, not the ceiling, and the ceiling is the media literacy curriculum the procurement budget does not fund.
The Sunday thread is that the three reframes the sector landed this week (one on assessment design, one on what universities should test, one on what schools should teach about synthetic media) are different answers to the same question: what does the educator do when the technology moves faster than the policy? More on the operational read at digitalattitudes.com.au.