Where Norway Drew The Line

On June 23, Norway's prime minister told his country that when its youngest students return to classrooms in August, they will not be using generative AI in school.¹ No ChatGPT. No Gemini. No Copilot. Not on school devices, not for school work. The rule covers grades one through seven, the years from age six to thirteen. Older students get a graduated leash: fourteen to sixteen only under a teacher's supervision, seventeen and up trusted to work with the tools directly.² It is the first age-specific AI restriction in European schooling.³

The policy will be discussed in headlines for what it forbids. What is worth reading it for is the sentence Jonas Gahr Støre gave as the reason.

The Sentence Under The Policy

He said it plainly. The most important thing in school, he told the country, is that our children learn to read, write and do mathematics.¹ That is a sentence any education minister could have said in any decade. What makes it striking in 2026 is that it is being used, apparently for the first time by a national government, as the argument for keeping a technology out of a classroom that is currently being pushed into every other classroom in the developed world.

The Norwegian argument is not that AI is dangerous, or that the models are inaccurate, or that student data will leak. Those arguments exist in other countries. The Norwegian argument is smaller and older: that a six-year-old learning to sound out a word, and a nine-year-old learning to hold a paragraph in their head long enough to write the next sentence, and an eleven-year-old learning to sit with a division problem for the third minute rather than the third second, are doing work that only they can do. If a tool does that work for them, or beside them, or slightly ahead of them, it does not accelerate the learning. It replaces the person the learning was meant to build.

The Precedent Nobody In America Talked About

There is a reason Norway feels comfortable making this call. In 2024, the country banned smartphones in its middle schools. A study of more than four hundred of those schools, led by Sara Abrahamsson at the Norwegian Institute of Public Health, tracked what happened next.⁴ Reported bullying dropped by about forty six percent among girls and about forty three percent among boys.⁴ Specialist visits for psychological distress fell. Grade point averages rose, most for girls from lower income households, whose likelihood of attending an academic high school track went up alongside their marks.⁵

The country ran the experiment. It got the answer. It kept the rule. The AI policy is being written by people who watched that data land. They are, in effect, running the same experiment again with a new tool, and betting the results will follow the same shape. A generation that reads, writes, and does math on paper first, and only later meets the machine, will end up in a different place than a generation that meets the machine at six.

What The Machine Is Not Asked To Do

There is a version of the future of education that treats every question a student has as a question the machine should answer. The machine gets faster, the answers get better, the student gets more of them per hour. Progress, on that story, is measured in how quickly the friction disappears.

Norway is describing a different future. In that future, the friction is not a bug. It is where the child is actually forming. The pause before the sentence is not a delay to be optimized away. The paragraph deleted and rewritten in different words is not evidence that the tool failed. The third minute on the division problem is not a signal that the software should have intervened. Those are the moments the person is being made.

None of this requires nostalgia. A country can hold both truths at once. That at seventeen a student should know how to work with a large language model on serious work. And that at seven the same student should be left alone with a book and a pencil long enough to become someone who can. The tools are not the enemy. The order is not incidental.

At Koan we think about the same layer from the other side. Once a student is old enough to work with the machine, the question shifts. It is no longer whether the tool is present. It is whether the trace it leaves behind lets a teacher see the student's own thinking still happening underneath. The pause. The revision. The paragraph rebuilt in different words. The moment the student chose one framing and can, later, explain why. Norway's policy protects that layer by keeping the tool out until the layer has formed. Our work is to protect it after the tool arrives.

If the six-year-old in your building learned to read this year without ever asking a machine what a word says, and the seventeen-year-old learned to write this year with a machine at their side, which one will have become the more careful thinker at twenty five?