AI in Schools
Anthropic accuses Alibaba over Claude as ESA Module 2 ships
Digital Attitudes, 26 June 2026
The Friday read is one Anthropic letter to the Senate Banking Committee that names Alibaba's Qwen AI lab as the operator behind the largest known distillation campaign against Claude, one Sydney based finance academic's argument in Future Campus that Australian universities are auditing the wrong question on assessment in the AI era, and one Education Services Australia and Microsoft training release that puts the second of two free AI Implementation modules into every Australian teacher's professional learning library this month.
Anthropic names Alibaba's Qwen lab for harvesting Claude through 25,000 fake accounts
CNBC reported on Tuesday 24 June that Anthropic disclosed in a 10 June letter to Senate Banking Committee chair Tim Scott and ranking member Elizabeth Warren that operators linked to Alibaba's Qwen AI lab ran 28.8 million exchanges with Claude through approximately 25,000 fraudulent accounts between 22 April and 5 June 2026, with the campaign targeting Claude's agentic reasoning, software engineering proficiency and long horizon task completion capabilities. Anthropic characterised the operation as "the largest known distillation attack on Anthropic to date" and urged Washington to pursue stricter semiconductor export controls and pass laws penalising AI developers caught running distillation operations. The letter notes Alibaba pressed on after a Trump White House April memo put foreign entities on notice about industrial scale efforts to harvest US AI systems, and that Alibaba declined to comment by reporting time. CNBC on the Anthropic letter and the Alibaba distillation campaign, 24 June.
Why this matters: For Australian university Chief Information Security Officers and IT leaders signing or extending Anthropic enterprise contracts against the Cogniti Microsoft Marketplace listing the Wednesday brief carried and the UNSW ChatGPT Edu deal that sat 10,000 licences on Azure last September, the operational read is that the same vendor whose Mythos model the Trump administration ordered off the shelf for foreign nationals two weeks ago has now publicly disclosed that its commercial API was the access channel for the largest scraping campaign against a US frontier model on record, and the question the next academic board AI procurement paper answers is whether the standard Claude contract clause covers identity verification at the account creation layer or whether the institution's own single sign-on federation now sits in front of the API as the next compliance control. For state Department of Education AI procurement teams running the NSWEduChat, Corella, ClassmAIte and Victorian Government AI vendor lists against the AI Review Committee bar the Wednesday brief named, the read is that the federal panel has its first real test case in whether Anthropic's transparency on the distillation campaign clears the high residual risk panel's terms of reference, and the question the term 4 paper answers is whether the approved AI vendor list still sits Anthropic in the same tier as OpenAI and Google given the open account creation question the letter just put on the desk. For ed-tech vendors selling AI tutoring, AI marking and AI lesson planning products into Australian schools and universities (the ChatGPT Edu line UNSW signed with OpenAI, the Cogniti agents Sydney built on Azure OpenAI, the Khanmigo and MagicSchool products on the school side, the Pearson Writing Coach and Brightpath products on the assessment side), the procurement read is that every commercial AI vendor's account creation and identity verification posture is now the first question the school's chief information officer asks before the next renewal, and the vendor whose pitch deck cannot answer it is the one whose next contract conversation answers a buyer reading the Anthropic letter on the same tab.
Future Campus says audit the verification, not the production
Pedram Nourani, Assistant Professor of Finance at S P Jain School of Global Management in Sydney, argued in Future Campus on Monday 23 June that Australian universities are still using assessment built to measure student production when the skill the AI era actually rewards is human review of AI output. Nourani's frame: "The skill that survives, and grows more valuable, is review: directing the work, interrogating it, and owning the result," with the structural warning that "when output is usually right and always polished, vigilance decays and we wave it through." His proposed redesign shifts assessment toward tasks that ask the student to identify flaws in AI generated analysis rather than grading polished student produced work, while keeping production competence as the foundation the review skill rests on. Future Campus on the audit the verification not the production argument, 23 June.
Why this matters: For Pro Vice-Chancellors (Education) writing the 2027 assessment plan under TEQSA's regulatory led framework that started this year and the three pathways the September 2025 enacting assessment reform paper named (program wide reform, unit level assurance, hybrid model), Nourani's piece reads as the cleanest single sector statement available on what the unit level assurance pathway actually has to deliver, and the question the next academic board paper answers is whether the institution's existing AI detection contract (Turnitin's AI detector, the AI integrity work the institution sourced through Cadmus, the Pearson and Inspera proctoring stacks) gets re-priced against an assessment design that asks the student to verify AI output rather than against a tool that asks the system to detect AI input. For school principals reading the Liu two lane Sydney Assessment Framework the brief carried earlier this month and the Nourani audit the verify argument back to back against the term 4 reporting calendar, the read is that the assessment redesign conversation now has two cleanly stated lanes (the secure in-person exam the AI cannot enter, the verification task the AI cannot finish) and the question the term 4 staff meeting answers is whether the senior secondary marking rubric gets re-written this term or whether the existing rubric carries the year. For ed-tech vendors selling assessment, AI detection, plagiarism and academic integrity products into Australian schools and universities (Turnitin, Cadmus, Inspera, Pearson Writing Coach, the Brightpath and Mark My Words products on the K-12 side), the procurement read is that the buyer's frame is moving from "detect the AI" to "assess the verify" inside one budget cycle, and the vendor whose product cannot show what the verification competency looks like in the marking rubric is the one whose next renewal gets the conversation the academic board paper just printed.
ESA and Microsoft ship the second free AI Implementation module for Australian teachers
Education Services Australia and Microsoft Australia made the second module of the joint AI for Educators training pathway, "AI Implementation: Strategies for guiding students in safe and effective use of GenAI," available to Australian teachers through ESA's Digital Technologies Hub this month, with the module running approximately 90 minutes alongside the existing "AI Readiness" Module 1 and issuing a certificate participants can map against the Australian Professional Standards for Teachers and the Australian Government's Framework for Generative AI in Schools. Module 2 is designed for teachers of students aged 13 and above, builds on the foundation Module 1 set on evaluating AI generated content for bias and misinformation, and covers practical age appropriate classroom activities for guiding student GenAI use through worked examples and live online training sessions, with the training also counting toward Microsoft's commitment to skill 1 million people across Australia and New Zealand in GenAI by the end of 2026. Cyber Daily on the Microsoft and Education Services Australia AI training partnership and the staged module release plan.
Why this matters: For state Department of Education and Catholic system professional learning leads writing the term 3 PD calendar against the existing AITSL accredited microcredentials list, the operational read is that Module 2 is the artefact every teacher of secondary year levels can complete this term as a 90 minute free professional learning entry that maps cleanly to the Australian Professional Standards for Teachers, and the question the next staff meeting answers is whether the school adds Module 2 to the term 3 standing PD line or whether the existing AI literacy provider (the AISWA Nurture training, the Brisbane Catholic Education Copilot pathway, the Education Perfect AI courses) keeps the slot. For independent and Catholic school heads pricing the next PD investment against the Microsoft AI Literacy for Educators credential the Wednesday brief named and the existing Anthropic, Google and OpenAI vendor relationships, the read is that the ESA module is the public domain free credential that sits on top of the federal framework while the Microsoft credential is the private credential that sits on top of the vendor's product, and the question the term 4 PD paper answers is whether the school accredits both as the staff facing pathway or whether it picks the one that maps to the existing curriculum cycle. For ed-tech vendors selling AI literacy professional learning, AI marking and AI lesson planning products into Australian schools (Education Perfect, Nurture, MagicSchool, Khanmigo, Eduaide, the Pivot Professional Learning AI courses, the Atomi AI features), the procurement read is that the free ESA Module 1 and Module 2 pathway is now the floor the paid product gets compared against in the school's PD budget conversation, and the vendor whose offering cannot show what the paid product gives the teacher that the free module does not is the one whose next renewal carries the question the principal is reading from the ESA hub on the same tab.
The Friday thread is the Anthropic letter to the Senate Banking Committee that names a Chinese lab as the operator behind the largest distillation campaign against Claude on record, the Nourani piece in Future Campus that puts the audit the verification not the production frame on the assessment redesign desk, and the ESA and Microsoft Module 2 release that drops the second free AI training credential into every Australian teacher's professional learning library this month. More on the operational read at digitalattitudes.com.au.