1. 250+ Experts Demand a Five-Year AI Moratorium in Schools — and They’re Not Backing Down

More than 250 educators, researchers, and child development specialists signed an open letter demanding a five-year moratorium on AI use in primary and secondary education. The signatories include developmental psychologists, neuroscientists, and classroom teachers — the people who actually work with children, not the people who sell AI products to schools.

The letter argues that children’s developing brains are uniquely vulnerable to AI’s influence — not just through cheating or over-reliance, but through fundamental changes to how children learn to think, socialise, and regulate their own cognition. The proposed moratorium would cover all AI tools in K-12 classrooms, including “educational” chatbots, AI writing assistants, and personalised learning platforms.

Why it matters: This isn’t Luddite panic. These are the people who study child development for a living. Their argument — that we’re conducting an uncontrolled experiment on children’s cognitive development — is hard to dismiss. The question isn’t “can AI help kids learn.” It’s “what does AI do to the process of learning when used from age 5?” We genuinely don’t know the answer, and 250 specialists just told us we shouldn’t be finding out through mass deployment.

2. 31 States, 134 Bills: AI Education Regulation Is No Longer Guidance — It’s the Law

Across the United States, 31 states have introduced 134 bills addressing AI in education in 2026 alone — a dramatic acceleration from 2025’s 90 bills across 21 states. The legislation covers everything from mandatory AI literacy curricula to outright bans on AI use in certain grade levels and subjects.

Notable trends: 14 states now require schools to disclose when AI tools are being used in instruction. 8 states have banned AI writing assistants for students under 14. 6 states are piloting AI proctoring systems for standardised tests. The patchwork is real — but the direction of travel is clear: states are acting because the federal government isn’t.

Why it matters: The US is building AI education policy state by state, which means 31 different experiments in what works and what doesn’t. For New Zealand, which has no national AI in education policy, watching the US state-level data will be invaluable — we can learn from 31 approaches without running 31 experiments ourselves.

3. Alpha Schools Expands to More Cities — $55K AI-Only Learning Model Draws Growing Questions

Alpha Schools, the private AI-only education provider, is expanding to additional major cities despite mounting ethical questions about its model. At $55,000 per student per year, Alpha’s schools replace traditional teachers with AI-guided learning platforms — students interact primarily with AI tutors, with human facilitators serving a supervisory rather than instructional role.

Proponents argue students progress faster without the constraints of traditional pacing. Critics — including the 250+ signatories from item #1 — say the model is the exact uncontrolled experiment the moratorium is designed to prevent.

Why it matters: If the $55K price tag seems like a barrier, you’re missing the point. Alpha Schools are building the blueprint for AI-only education at the high end of the market. If it works — or even if it appears to work — the model gets replicated at lower price points. Within a decade, “AI teacher” could be the default in private education, and “human teacher” becomes the premium option. The ethical conversation needs to happen now, while the stakes are high but not yet locked in.

4. “Pro-Human” AI Medical Training Tool Launches in NZ — A Different Philosophy for Education AI

A New Zealand-built AI medical training tool has launched with an explicit “pro-human” philosophy — designed to assist learning rather than replace teaching. The tool simulates patient interactions, lets learners make mistakes safely, and provides detailed feedback on clinical and communication skills.

The design is notable for what it doesn’t do: it doesn’t pretend to be a real doctor, it doesn’t grade students, and it doesn’t feed data back to an administration system. It’s a training sandbox, not an assessment tool.

Why it matters: This is the right way to build AI for education. It’s transparent about being AI. It’s designed to augment learning, not evaluate it. And it’s built by a Kiwi team who understand the NZ context — including cultural safety in clinical training. More of this, less of “replace the teacher.”

5. Two Studies Confirm What Teachers Suspected: AI Studying Reduces Retention and Social Competence

Two peer-reviewed studies published this week add to the mounting evidence that AI use in education has serious downsides. The first study found that students who use AI to study remember significantly less than those who don’t — and critically, they don’t realise they’re learning less. The second found the first peer-reviewed evidence that AI dependence is making students socially incompetent, with reduced interpersonal skills and emotional awareness.

Together, the studies paint a concerning picture: AI makes students feel like they’re learning more while they’re actually learning less, and it’s impairing the social skills they need to function in human workplaces. The gap between “AI gives better answers” and “AI teaches better” is wider than anyone in edtech wants to admit.

Why it matters: If these findings hold up at scale, they have massive implications. Students who rely on AI through their education will graduate knowing less and relating to people worse than the generations before them. That’s not a technology problem — that’s a civilisation problem. The answer isn’t “ban AI.” The answer is “teach with it consciously, not as a crutch.”

🔍 THE BOTTOM LINE

Education is the one domain where the AI industry’s “move fast and break things” philosophy collides directly with children’s development, and the evidence is mounting that we are breaking things. The 250-experts open letter, the two peer-reviewed studies on retention and social competence, and Alpha Schools’ expansion all point in the same direction: we rushed AI into classrooms without understanding the consequences. The good news is that the conversation is finally shifting from “how do we deploy AI” to “should we deploy AI” — and the answer is “not everywhere, not for everything, not yet.”