This week's roundup is dominated by pushback: New York City has just enacted the country's broadest school AI ban, and NSW has moved to limit take-home assessments after a wave of academic integrity cases. There's also fresh data on the training gap facing teachers, and a useful reminder that giving students access to an AI tool doesn't guarantee it gets used well.
From Term 4 2026, NSW schools can set no more than one unsupervised take-home HSC assessment task, worth a maximum of 15% of the school-based mark, after the Department asked NESA to urgently review AI's impact on assessment integrity.
Why it matters: If you're setting take-home tasks in a NSW secondary school, your assessment program needs auditing now — and the same "how do we know this is the student's own work" question applies just as much to homework set in primary classrooms.
NSW is restricting unsupervised assessments after more than 1,000 recorded cheating cases in 2025, South Australia has rolled its own secure "EdChat" tool out to high schools, and Victoria is introducing daily screen-time caps for primary students from 2027.
Why it matters: There's no single national approach yet, so what's allowed or restricted for your students depends heavily on your state and sector — worth checking your own department's current position rather than assuming it matches a neighbouring system.
New York City Public Schools — the largest US school district, serving nearly 600,000 students — has imposed a one-year moratorium on student-facing generative AI for grades 2K–8, alongside new screen-time limits and twice-yearly AI critical-thinking modules for high schoolers.
Why it matters: It's the biggest test case yet of a "build capacity before granting access" approach for younger students, and the district's year-long review will shape the evidence other systems point to next.
Following bans in New York City and Los Angeles, a growing number of parents, teachers and even school-board candidates in cities like Chicago are campaigning for moratoriums on generative AI, arguing schools moved faster than the evidence justified.
Why it matters: It's a useful reality check if your own planning has assumed AI adoption only moves in one direction — community buy-in, or the lack of it, is now a live variable in any rollout.
A Morning Consult survey of just over 2,000 US educators and parents found 76% of middle-school teachers use AI weekly, yet only 20% have received substantial training, and just 24% of educators think the system is adapting well to AI.
Why it matters: It's a concrete data point to bring to a leadership conversation about PD budgets — the gap here isn't interest or access, it's structured training time.
More than 25 education ministers meeting at UNESCO's Digital Learning Week in Paris adopted a joint statement on "sustaining education as a common good in the age of AI," alongside new policy briefs on AI safety governance, procurement and age-appropriateness.
Why it matters: The accompanying policy briefs are a useful reference if you're drafting or reviewing a school AI policy and want language that's been vetted internationally rather than written from scratch.
Stanford National Student Support Accelerator · 2026
A randomised trial of 355 students in grades 1–5 across two US districts found that simply providing access to an AI literacy tutor produced very low usage — as little as two minutes a week — and pairing it with human tutor support only modestly improved engagement.
Why it matters: If you're considering an AI tool for younger students, this is a reminder that access and a good product aren't enough — it's the implementation and adult support around a tool that determines whether it actually gets used.
A five-minute read for busy teachers: the week's most useful AI-in-education news, Australian stories first, each with a note on why it matters in your classroom. Free, and you can unsubscribe at any time.
