Children have seized on artificial intelligence with startling speed. UNICEF estimates at least 20 million of them across 10 countries have already used the technology. Many adopt it more than three times faster than adults. The findings come from fresh analysis released at the end of June. They paint a picture of young users turning to chatbots for homework help and personal guidance alike.
The numbers stand out. An estimated 13 million children rely on AI to support their learning and school assignments. More than 2 million, or one in 10 of those surveyed, seek its counsel on matters that worry them. These patterns emerge from nationally representative surveys of internet-using youth aged 12 to 17 and their parents or caregivers. The countries span diverse regions: Armenia, Brazil, Colombia, the Dominican Republic, Jordan, Mexico, Montenegro, North Macedonia, Pakistan and Serbia. Researchers weighted the results against population data to reach the global projections.
UNICEF laid out the data in a statement timed for the first Global Dialogue on AI Governance. “AI is here,” the agency declared. “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 organization stressed that new evidence reveals both the scale of uptake and the risks and divides that accompany it.
Yet the enthusiasm carries sharp edges. One-third of the children voiced worries that AI could be used to scam or trick people, or to spread misinformation. A quarter expressed fear that their own images or videos might be manipulated into sexually explicit deepfakes. Safety appears an afterthought in too many systems that reach young users. Children find themselves more exposed to how these tools are designed, how their data gets used, and the business models behind them. They possess far less power to push back.
Evidence on longer-term consequences remains thin. Impacts on cognitive development, emotional dependency and vulnerability to harm are only beginning to surface. A generation, in UNICEF’s words, grows up inside a global experiment. The agency paired the adoption snapshot with a separate policy brief titled “When AI becomes a friend.” That document examines child rights risks from AI chatbots and companions. It notes young people lean on these systems not just for facts or creativity but for advice, support and sometimes relationships.
The brief, released in June, calls for a shift. Regulators must move from reactive fixes to preventive strategies. An ecosystem approach should assign clear roles to governments, companies, parents, educators and communities. Few jurisdictions have crafted chatbot-specific rules so far, even as recognition grows that conversational AI poses distinct dangers for minors. “There is an urgent need to shift regulatory approaches from reactive to preventive,” the report states, “leveraging an ecosystem approach with roles and responsibilities for a range of stakeholder groups.”
Recent coverage amplified the alarm. Euronews reported on July 1 that safeguards for young users trail badly. The story highlighted how millions turn to AI for homework and, in some cases, personal advice. It echoed UNICEF’s concern that governance frameworks rarely put children first. Those who will live longest with the technology feel its weak spots soonest.
Broader data on youth and AI point in the same direction, though they vary by region and age slice. A Gallup survey from April found 51 percent of Americans aged 14 to 29 use AI daily or weekly. Gen Z students in K-12 settings reported even higher weekly engagement than their slightly older peers. Other analyses show younger adults under 30 lead in tasks such as brainstorming and information searches. Adoption among teens in certain U.S. studies reaches one-third overall, climbing near 50 percent for ages 15 to 17. These figures reinforce the sense that digital natives integrate the tools faster than their elders.
But speed does not equal safety. UNICEF’s snapshot draws from the Disrupting Harm Phase 2 project, a collaboration with ECPAT International and INTERPOL funded by Safe Online. Roughly 1,000 children and 1,000 parents or caregivers were polled per country. The design aimed for near-full national coverage. The resulting estimates carry weight. They arrive as policymakers gather to debate global AI rules.
The agency offers a clear list of demands. Governments, industry and partners should embed child rights—especially the right to safety and protection—into every layer of AI governance. That means investing in research on how the technology affects development and well-being, with emphasis on hazards. Laws and corporate accountability must tighten to prevent AI-enabled sexual exploitation and abuse. Systems need to be built with maximum safety and transparency from the start. AI literacy programs should reach children and the adults who guide them. And digital infrastructure must expand so no child is left on the wrong side of the divide.
These steps matter. The choices made today will echo for decades in children’s privacy, well-being and access to opportunity. Companies that develop or deploy AI across its life cycle receive tailored recommendations in UNICEF’s accompanying business guidance. The advice draws on international best practices without pretending to serve as a compliance checklist. Developers and deployers face different responsibilities. Both must prioritize prevention.
Industry insiders have watched generative AI permeate education and daily life at a clip few predicted. Teachers experiment with it for grading and feedback. Students summon it for essays and explanations. The UNICEF data adds urgency to questions of oversight. When one in 10 young people consults AI about their deepest concerns, the stakes rise. Emotional dependency on machines that lack true understanding could reshape relationships and self-perception in ways still hard to measure.
Misinformation risks feel immediate to the children themselves. So do fears of manipulated media. Deepfakes that weaponize a young person’s likeness threaten reputation, mental health and physical safety. Current platforms often ship without age-appropriate defaults or robust reporting tools. Many lack even basic transparency about how they handle children’s data or moderate outputs.
Connectivity gaps compound the problem. Not every child enjoys reliable internet at home or school. Those who do may encounter tools optimized for engagement rather than protection. The result is an uneven experiment. Some gain creative sparks or personalized tutoring. Others stumble into unreliable advice or harmful content. The digital divide, once about access alone, now includes differences in safe and informed use.
UNICEF’s timing feels deliberate. The Global Dialogue on AI Governance offers a forum to push child-centered principles into international standards. Past technology waves—social media, smartphones—taught hard lessons about bolting safeguards on after the fact. Regulators and companies cannot repeat the pattern.
Research must accelerate. Longitudinal studies on cognitive and emotional effects deserve funding. Developers should test rigorously with child users before wide release. Policymakers need to demand age assurance, default privacy settings and clear accountability when harms occur. Parents and teachers require practical guidance to talk about AI with the young people in their care.
The alternative looks messy. A generation internalizes patterns set by profit-driven algorithms. They learn to outsource thinking, to trust outputs that blend truth and fabrication, to seek companionship from systems incapable of genuine care. None of that aligns with stated goals of fostering critical minds and resilient hearts.
Progress is possible. Some firms already experiment with child-specific models that cite sources, flag uncertainty and refuse inappropriate requests. Schools pilot curricula that teach prompt engineering alongside media literacy. Governments in parts of Europe and elsewhere draft rules that explicitly reference minors. Yet scale remains limited. The UNICEF figures show how quickly the ground shifts underneath these efforts.
Twenty million children. Three times the adult adoption rate. Thirteen million turning to AI for homework. Two million asking it about their fears. The statistics are not abstractions. They represent real boys and girls whose daily experiences now include conversations with machines. The adults charged with their protection cannot afford to move slowly.
The moment demands action that matches the pace of uptake. Research. Regulation. Design choices that put safety first. Literacy at every level. Infrastructure that reaches everyone. These elements form the foundation for AI that augments childhood rather than gambles with it. The data from 10 countries delivers an early warning. Heed it now, before the experiment writes its own unchecked conclusion.
WebProNews is an iEntry Publication