More than 20 million children worldwide have already used artificial intelligence tools. They pick them up three times faster than adults do. And the numbers keep climbing.
UNICEF laid out the scale in a statement released June 30, 2026. The agency drew from surveys across 10 countries. Roughly 13 million children turn to these systems for schoolwork and homework. More than 2 million, or about one in 10, ask AI for advice on personal worries.
The findings arrive just as governments gather for the first Global Dialogue on AI Governance in Geneva. UNICEF wants child rights placed at the center of those talks. Without that focus, the organization warns, a generation grows up inside an uncontrolled experiment.
Children recognize the dangers. One-third worry that AI will help scammers trick people or spread false information. One-quarter fear their own photos or videos could become sexually explicit deepfakes. Those fears sit alongside genuine excitement about new ways to learn and create.
Yet evidence on how these tools affect young brains, emotions and social skills remains thin. Cognitive development. Emotional dependency. Exposure to harm. All of it stays largely uncharted. “AI is here. It is a growing part of all of our lives. And it is already shaping childhood around the world – for better and for worse,” the UNICEF statement declares.
The agency points to uneven access. Kids in wealthier households or urban areas dive in first. Those in the global South often lack reliable internet or devices. The result? A widening gap that leaves some children behind while others race ahead without proper safeguards.
Separate research tied to the same Disrupting Harm project delivers even sharper warnings. At least 1.2 million children reported that someone had manipulated their images into sexually explicit deepfakes in the past year. In certain countries the rate hit one in 25 – roughly one child in every classroom. Forbes reported on the UNICEF-backed findings in February 2026.
Perpetrators no longer need advanced skills or expensive equipment. Open-source models run on ordinary laptops. What once demanded expertise now takes minutes. “Deepfake abuse is abuse, and there is nothing fake about the harm it causes,” UNICEF stated. The shame, stigma and isolation hit victims hard. Even fabricated images fuel demand and normalize sexualization of children.
Parents and educators see the shift too. A Harvard Graduate School of Education podcast last year explored similar territory. Ying Xu, who studies the topic, noted that many children treat AI companions like people. They share private details. They assume the systems understand them. Early data suggests this blurring can shape how kids learn to form relationships or evaluate truth. The Harvard EdCast episode from October 2024 highlighted the urgent need for AI literacy programs that teach children to question outputs and spot manipulation.
Schools stand at the front lines. Some systems already integrate generative tools for personalized tutoring. Others ban them outright. A Brookings Institution analysis released in January 2026 argued that current risks to cognitive growth outweigh benefits in many educational settings. Students may skip the hard work of reasoning when AI supplies answers instantly. Foundational skills erode. NPR covered the Brookings report in January 2026.
UNICEF lays out a clear list of steps. Governments must invest in research that tracks long-term effects on development and well-being. Laws need updating to cover AI-enabled sexual exploitation with real penalties. Companies should build systems with safety and transparency as defaults, not afterthoughts. Digital literacy programs must reach children and caregivers alike. And countries have to expand connectivity so the AI divide does not become permanent.
These recommendations echo in UNICEF’s updated policy guidance released late last year. Version 3.0 stresses that AI policies must uphold child rights at every stage of design and deployment. The document offers checklists for lawmakers and developers. It insists on meaningful consultation with young people themselves.
Recent discussions on X show the conversation spreading. UNICEF Montenegro posted this week that nearly half of internet-using children aged 12 to 17 in the country have already tried AI platforms. The agency called for literacy to become part of national education reforms. Similar posts from other UNICEF offices highlight local surveys that mirror the global numbers.
But agreement on solutions remains elusive. Tech firms push for innovation-first rules. Regulators worry about overreach. Child advocates demand guardrails now. The Geneva dialogue this week offers a chance to align those views. Whether children’s voices cut through stays an open question.
The stakes stretch decades into the future. Today’s middle schoolers will enter adulthood alongside far more powerful systems. Their privacy, mental health and economic opportunities could hinge on decisions made in conference rooms far from any classroom. Safety cannot remain an afterthought. Research cannot wait. Access cannot stay unequal.
UNICEF puts it plainly. The choices made about AI right now will shape children’s safety, privacy, well-being and equal chances for years ahead. The data shows the adoption curve bending sharply upward. The question is whether policy and protection will finally catch up.
WebProNews is an iEntry Publication