The canteen must prove it is safe. The software does not have to.

The story

Schools are adopting educational technology before knowing what damage it could do to children, experts have warned. A major review seen by Metro concludes that too little is known about whether digital learning tools such as AI and VR harm children's wellbeing and development. Researchers at the LSE-backed Digital Futures for Children Centre found that more than 2,100 studies of educational technology overlooked questions about safety and focused instead on test scores. Campaigners are calling on education secretary Lucy Powell to address the technology before it takes over classrooms. Colette Collins-Walsh of the 5Rights Foundation, which backed the report, said school canteens cannot serve food unless it is proven to be safe and beneficial to children's healthy development, yet digital products have been introduced into classrooms without being held to the same standard, and that it is time tech had its "Turkey Twizzler moment", a reference to Jamie Oliver's campaign against processed food in schools. Daniel Kebede of the National Education Union said the government should take the same safety-first approach to potentially harmful edtech products as it does with social media.

The reframe

The comparison being made here is more useful than the campaigning language around it, so it is worth isolating. A school is a building into which two kinds of thing are introduced for children to consume. One of them, food, cannot enter unless it satisfies a nutritional standard. The other, software, can enter on the basis of a procurement decision and a supplier's own claims. Both are consumed daily by the same children, in the same rooms, under the same duty of care.

The point is not that food regulation is excessive. It is that the existence of food regulation settles the question of whether this kind of standard is possible. Schools already operate a precautionary rule, already administer it, already absorb its costs, and already refuse things that fail it. The capacity is proven. It is simply not extended.

What distinguishes the two categories is not evidence of harm, because on the review's own findings the evidence about digital tools has barely been gathered. It is not urgency, since children are using these products now. The distinguishing feature is that food in schools became a public scandal with a face attached to it, and educational software has not. The Turkey Twizzler reference is more precise than it first appears. The standard that now governs school food did not emerge from a considered assessment of what children consume. It emerged from a television campaign that made one product visible enough to be embarrassing.

The 2,100 studies are the part of this story that should give most pause. That is not an absence of research. It is a large body of research that asked a different question. Attainment was measured because attainment is what schools are held to account for, what suppliers can market against, and what generates a publishable result. Whether the tool harms the child was not measured, not because anyone decided it did not matter, but because nothing in the structure required anyone to ask. The metric that existed absorbed the effort available, and the question with no metric attached went uninvestigated for the length of an entire technology's adoption.

This is what the book means when it says that education transmits expectations about legitimacy rather than knowledge. The lesson available to a child in a classroom running an untested AI product, in a school where the same child's lunch has been assessed against a statutory standard, is not about either the lunch or the software. It is that the level of care an institution takes over what it gives you depends on which of its obligations happens to be written down. Adults learn that lesson too, usually later and at greater cost.

Nobody in this story is behaving badly. The schools are buying tools they are told will help. The suppliers are answering the questions they were asked. The researchers studied what the funders wanted studied. The government is applying a safety-first approach to social media, which is where the political attention currently sits. Every part of the arrangement is defensible on its own terms, and the result is that the one category of product entering a school with no proof-of-safety requirement is the one nobody has yet made a television programme about.