A not-for-profit effort of the Webspinner Foundation
Governments and school districts are restricting children's access to artificial intelligence. It is the modern equivalent of banning books — and it will work about as well.
AInoculate builds free, hands-on AI training for K‑12 classrooms, so the generation that will inherit this technology is equipped to govern it rather than merely survive it.
The situation, as of this week
On September 2, 2026, New York City announced the broadest student-facing generative AI moratorium in the nation: every student from 2‑K through eighth grade, for one year, with 38 previously approved programs discontinued.
We take the district's stated concerns seriously. Several of them are correct, and we say so plainly. Companion chatbots do not belong in the hands of minors. Answer-dispensing homework tools measurably damage learning. The evidence base is genuinely thin.
But the policy bans the wrong noun — and the children it is meant to protect are already, overwhelmingly, on the other side of the fence.
What a school-network ban does not reach
of U.S. teens use AI chatbots. About 30% use one every day.
Pew Research Center, Feb 2026 · n = 1,458
have used an AI companion — and only 37% of parents know their child uses AI at all.
Common Sense Media · n = 1,060
of teens who used AI on a school assignment did so without their teacher's permission.
Common Sense Media Census, 2026
children aged 12–17 reported having their images turned into sexual deepfakes in a single year, across 11 countries.
UNICEF · ECPAT International · INTERPOL
of identified school deepfake victims were girls — 600+ victims across ~90 schools in 28 countries since 2023.
WIRED / Indicator investigation, 2026
fake nude images of 59 classmates, made by two 14-year-old boys using photographs published by their own school.
Pennsylvania preparatory school case
Read that last one carefully. Those images were made by children, of children, on personal devices, outside school hours, from school-published photographs. A moratorium on district networks would not have prevented a single one of them. What might have is a sixth-grade unit on synthetic media, provenance and consent — which is precisely the instruction a K–8 moratorium makes difficult to deliver.
The Inoculation Thesis
Prohibiting generative AI in K–12 does not remove AI from childhood. It removes teachers from AI.
Exposure continues at the same rate, in the same volume, in the same products — minus the one adult in the child's life who was paid to be in the room, who knew what a hallucination was, and who could say check that.
And it does not remove teachers evenly. The child with a subscription at home and a parent in technology loses nothing. The child in one of the 15.7 million American households without high-speed broadband loses the only supervised, expert-mediated encounter with AI they were ever going to get. A ban converts AI literacy from a public capability granted by schools into a private advantage purchased by families. That is the definition of a regressive policy.
The child hands the machine their work, and the machine does it.
The child is handed the machine's output and asked to find the error, name the technique, trace the source, prove the claim. The machine does no work; it is the work.
A moratorium that does not distinguish these forbids the second in order to stop the first. That is the error — and it is correctable. It requires no retreat from the district's safety commitments. It requires a definition. Read the full argument →
The programme
Inoculation is not a metaphor we chose for the branding. It is a specific psychological mechanism with forty years of experimental support: exposure to a weakened form of a manipulation, delivered with refutation, confers durable resistance to stronger versions later.
Every dose is free, open-licensed, offline-capable, and collects no student data.
Unplugged. No screens, no student accounts. One idea: a machine that guesses the next word is not a person, and does not know things. The class collectively plays the model, and discovers it can produce a confident ending without knowing whether it is true.
Teacher-projected, whole class. Hallucination hunts on the class's own subject matter, a year-long "Confident Wrong" wall of specimens, and the check-it reflex drilled to automaticity — lateral reading, three sources, always.
The crisis dose. Provenance and forensics, synthetic media and consent, voice-cloning scams, and a concrete response protocol for someone made a fake of me, or of my friend. Delivered in the grade band where peer image abuse begins — the band a K–8 moratorium places out of reach.
You resist manipulation by learning to perform it. Students build a weak manipulator in a sandbox to learn its moves from the inside, then replicate the MIT cognitive-debt finding on themselves and recover the guardrail result with their own data.
These systems will be governed by people. You are those people. Red-teaming, model auditing, and impact assessment of real proposed deployments — in a hospital, a court, a hiring pipeline, a school. Students draft an AI acceptable-use policy for their own school and defend it before a real school board. Case studies in whistleblowing, conscientious objection, and the practical mechanics of saying no.
Why we are confident this will not work
We are not speculating about what happens when schools respond to a frightening new influence by restricting access to it. There is a record.
| The restriction | What it was meant to do | What happened |
|---|---|---|
| Abstinence-only sex education | Prevent teen pregnancy by withholding information | CDC meta-analysis of 66 comprehensive vs. 23 abstinence programs: comprehensive programs improved outcomes; abstinence-only showed insufficient evidence of any change. State-level policy correlates with higher teen pregnancy. |
| Book bans | Protect children from harmful ideas | ~23,000 bans since 2021; 6,870 in 2024–25 alone. 79% of banned titles were written for children and young adults. Restricting access reduces reading volume — among the strongest predictors of achievement. |
| CIPA internet filtering | Keep students away from harmful content | Two decades of over-blocking legitimate health and LGBTQ+ material, while technically fluent adolescents route around it. Constrains the compliant student; fails against the one it was written for. |
| Cellphone bans strongest counter-case |
Restore attention and raise achievement | The largest U.S. study: in-class phone use fell from 61% to 13% and well-being improved — but average effects on test scores were near zero. A Florida study found gains after an adjustment year. The evidence genuinely splits. |
The honest summary: prohibition reliably reduces supervised use, and unreliably produces the learning outcome anyone actually wanted. That deserves respect as a finding. It is not an argument for banning the thing we are supposed to be teaching.
In the news
It would be easy to read this far and conclude the situation is grim. It is not. Across governments, institutions and student bodies, people are arriving independently at the same conclusion — that children are protected by being taught, and that AI literacy belongs in the curriculum rather than behind a fence.
We did not organise any of this. We are pointing at it because it is working, and because it is the direction worth joining.
Select a placard for the synopsis and the link to the original reporting.
Why this organisation exists
Someone will have to audit a model for discriminatory outcomes, refuse an unlawful deployment, write the statute, testify about the harm, build the countermeasure — and, when a synthetic voice calls in their grandmother's tone asking for money, recognise it.
Governance capacity is not a credential issued at twenty-two. It is a disposition, formed early: the reflex to ask who made this, what does it want from me, and how would I check?
You cannot raise a generation capable of governing a technology by arranging for their first serious encounter with it to be alone, at night, on a phone, with a product optimised for engagement rather than for truth.
We do not keep the genie in the bottle by refusing to look at it.