A coalition of 17 countries wants children to have a say in the AI that educates and monitors them

By Mirage News · July 8, 2026.

The Coalition for the Rights and Protection of Children in the Age of Artificial Intelligence was launched this week in Geneva, as part of the UN Global Dialogue on AI Governance. It is made up of six United Nations agencies —among them UNICEF, UNESCO, the ITU and the human rights office OHCHR— together with 17 governments (Austria, Brazil, Bulgaria, Canada, the Czech Republic, El Salvador, Estonia, France, Indonesia, Italy, Japan, Kenya, Luxembourg, Morocco, the Netherlands, South Korea and Spain) and the European Commission. Its starting point is the Convention on the Rights of the Child, the most widely ratified human rights treaty on the planet, and its central proposal is a shift in approach: minors should not be treated as users who are protected after the fact, but as rights holders whose voice must be incorporated from the very design of AI systems. The initiative comes after the UN Secretary-General's call for a 'Child Safety Pact for AI' at the dialogue's opening.

It is a declaration of intent, not a treaty. Members commit to sharing evidence and best practices and to pushing for children's opinions to 'genuinely inform' decisions about systems that affect them. There are no, at least in what has emerged from the launch, sanction mechanisms, mandatory audits or concrete deadlines. And what stands out, above all, is who is not on the list: neither the United States nor China, the two countries where practically all development of the large models these children will use daily is concentrated, have signed. Without the two AI superpowers, the coalition runs the risk of becoming a coordination forum among the regulated without regulating those who actually build the technology.

That said, the underlying diagnosis is accurate and honest about the short-term problem: today minors grow up inside systems —recommenders, AI tutors, conversational assistants— designed without them in mind, and safeguards tend to arrive after the harm, not before. That governments with very different regulatory approaches (from the EU to Indonesia, from Japan to Kenya) sit down to share a common framework based on an already established human rights treaty is, in itself, a relevant step: it provides a shared legal language —the 'right to be heard'— on which to hold companies accountable, even if they have not signed anything.

Our reading is that this type of coalition is the necessary, though insufficient, scaffolding to ensure that AI's long-term promise in education —personalized tutors, universal access to quality knowledge, early detection of learning difficulties— is not built on a generation of users whose data, attention and well-being were adjustment variables. The abundance that AI in education can bring only holds up if the infrastructure of trust is built now, with clear rules about which algorithms decide what a child sees and why. The short-term risk is that declarations like this remain a diplomatic gesture while the industry moves at another pace; the long-term opportunity is that they set a precedent so that, when regulatory pressure arrives in earnest —and it will, driven by parents, educators and teenagers themselves— there is already a reference framework with international legitimacy rather than starting from scratch country by country.