On 2 September, New York City banned generative AI outright for roughly 600,000 public-school students, pre-kindergarten through Grade 8, which is close to two-thirds of the system's entire enrolment. Mayor Zohran Mamdani framed the moratorium as protecting "human connection" in the classroom; whatever you make of the politics behind that phrase, the scale of the number is the thing worth sitting with, because a major Western school system, having spent two years telling teachers to experiment with AI, has just decided its youngest two-thirds should have none of it at all, for at least a year.
Four days later, two academics from Universiti Utara Malaysia's School of Education, Associate Professor Dr Muhammad Noor Abdul Aziz and Dr Siti Noor Aneeis, used that same ban as the hook for an opinion piece in Malay Mail, arguing for what they called an age-progressive AI curriculum for Malaysian schools: primary pupils build the underlying skill first, writing a paragraph, solving a problem, before AI is allowed anywhere near the task, and secondary students study AI as a subject in its own right, how it works, why it hallucinates, where bias enters, rather than simply being handed it as a tool. Their argument is more useful than the ban that prompted it, because they were careful, rightly, to treat New York as a provocation rather than a template. Cite the moratorium; don't copy it.
What New York actually did, and why a school here can't do the same
Read past the headline and the New York policy is more graduated than "ban" suggests. Chatbots are off-limits pre-K through Grade 8, correct, but high schoolers still take two AI critical-thinking classes this school year, and a small number of pilot classrooms, capped at five per school, keep AI access under a teacher's direct supervision. That is not a system that has decided AI has no place in education; it is a system that decided its youngest students were not yet ready for unsupervised access and built the exposure back in gradually from there. A Malaysian international school lifting the "ban" half of that policy and leaving the rest behind would be copying the part that solves a New York problem, an ungoverned rollout across a huge public system, while ignoring the part that actually transfers.
It also runs straight into a mandate NYC never had to worry about. The 2027 curriculum reform instructs Malaysian schools to build digital and AI literacy into the timetable from primary upward, through Teknologi dan Digital and the wider push toward Know, Understand, Do and, from Year 3, Apply as assessment tiers that increasingly assume some comfort working alongside these tools. A school that bans AI outright for its Year 1 to Year 6 pupils is not being cautious within that mandate; it is quietly opting out of it, and will have a harder conversation to have with parents and inspectors in 2027 than the one it is avoiding now.
Before your SLT writes its own AI-in-primary-years policy, don't start from an age cutoff. Start by naming the specific unassisted skill you want a pupil to demonstrate before AI is allowed near that task in that year group: composing a paragraph without a draft to edit, working through long division by hand, forming an opinion before seeing what a chatbot says. Name the skill, not the age. That is the decision New York's blanket cutoff skips, and the one Malaysia's own mandate expects your school, not a ministry circular, to make.
The principle worth stealing
Strip the politics and the headline out of the New York decision and what is left is a defensible idea: readiness, not age on its own, should govern when a student is handed an AI tool without supervision, and readiness has to be demonstrated, not assumed because a birthday passed. That is precisely the UUM academics' age-progressive argument, and it happens to fit Malaysian schools rather better than it fits a system the size of New York City's, because a single international school, with a few hundred pupils and a staff who actually know them, is in a far better position to judge individual readiness than a policy written for 600,000 children across a thousand-plus schools ever could be. The ban was the only lever New York had at that scale. Your school has better ones.
Why this matters for Malaysian international schools specifically
The honest answer is that neither New York's ban nor the ministry's still-unpublished syllabus detail settles, for your school, the question of when a Year 4 pupil should be allowed to ask an AI tool for help and when they should not. That judgment sits with your SLT, and the 2027 mandate makes writing it down, rather than leaving it to individual teacher discretion or a New York headline, the actual job of the next two terms.
During a recent visit to Australia I happened to watch a government official say seriously, and somewhat proudly, that Australia was leading the initiative to counter the negative effects of digital screens and social media with their ban in schools. I agree—letting any technology simply run free in society without pre-emptive care is not just irresponsible, but downright foolhardy. With AI, the effects can be even more catastrophic. And while we are pretty much building the plane while flying it, there is already awareness and talk of not only slowing down, but even "pulling the plug".
Want a second opinion on where your own AI-in-primary policy should land?
I run short diagnostic conversations with school leadership teams: not a sales pitch, a look at what age-progressive AI access would actually mean for your year groups, ahead of the 2027 syllabus detail.
Book a Discovery CallWhat This Means for You
New York's ban will be revisited in a year, as moratoria usually are, and by then some other jurisdiction will have made a different, equally headline-grabbing call. Chasing each one as it lands is not a policy. The principle underneath this one, readiness demonstrated before access granted, is durable enough to write into your own school's AI policy now, well ahead of whatever the ministry eventually publishes and long after this particular headline has been replaced by the next.