Children worldwide are embracing artificial intelligence at a pace that dramatically outstrips adult adoption, according to new data released by the United Nations Children's Fund. Drawing on research across 10 countries, UNICEF found that the rate of AI uptake among young people is more than three times that of their parents' generation, underscoring a fundamental shift in how the digital landscape is being experienced by the world's youth. This acceleration comes as AI systems become increasingly embedded in everyday applications, from educational platforms to social media, making exposure almost inevitable for connected children regardless of their digital literacy.
The scale of young people already engaging with AI technology is substantial and growing. UNICEF estimates that at least 20 million children globally have used some form of artificial intelligence, a figure that encompasses both deliberate adoption and incidental exposure through apps and services. More concerning is the subset seeking guidance from these systems: over two million children, representing roughly one in ten of all young AI users, are turning to artificial intelligence for advice on matters causing them personal anxiety or distress. This reliance on AI for emotional support and counsel reflects a troubling gap in traditional support networks and highlights how quickly these tools are becoming embedded in children's decision-making processes.
Educational deployment of AI represents another significant domain of childhood engagement. Approximately 13 million children across the surveyed nations are leveraging artificial intelligence to support their schoolwork and complete homework assignments. While this application offers legitimate pedagogical benefits—personalised learning paths, instant feedback, and accessibility improvements—it also means that educational outcomes for a substantial portion of the world's student population are now mediated through algorithmic systems whose design principles and business incentives may not align with children's actual learning needs.
The asymmetry between children's exposure to AI systems and their capacity to understand or resist them presents a critical vulnerability. Young people encounter these technologies in a state of relative powerlessness: they have minimal knowledge of how these systems are architected, which commercial interests drive their development, or how their personal data flows through corporate infrastructure. Unlike adults who may exercise consumer choice or regulatory advocacy, children lack both the cognitive frameworks to evaluate AI critically and the institutional power to demand accountability. This imbalance becomes particularly acute when considering that children will inherit the long-term consequences of governance decisions made today, yet have virtually no voice in determining those outcomes.
Specific harms already manifesting in children's AI experience have begun to emerge from UNICEF's research. One third of children surveyed in the 10 countries expressed anxiety about AI systems being weaponised for fraud, social manipulation, or the deliberate dissemination of false information. These concerns reflect real-world incidents where algorithmic content amplification has accelerated misinformation campaigns, often targeting younger users perceived as more credulous. Equally alarming is the vulnerability related to synthetic media: a quarter of surveyed children reported fear that their images or videos could be non-consensually manipulated into sexually explicit deepfakes, a threat that conflates AI technology with sexual exploitation and underscores the intersection of digital and physical harms.
The governance vacuum that UNICEF identifies extends across multiple institutional levels. Platforms and developers are reportedly releasing AI systems to child users with minimal protective infrastructure in place, suggesting that safety considerations remain peripheral to product design and deployment timelines. This pattern reflects a broader pattern in the technology sector where commercial velocity consistently outpaces safety evaluation, with children bearing disproportionate risk. The absence of guardrails means that the onus for protection falls on individual users and families rather than systemic design, an expectation that ignores vast disparities in digital literacy and access to protective resources.
For Southeast Asian nations, this global trend carries particular implications. The region's rapid digital adoption, combined with substantial youth populations and emerging regulatory frameworks, means that many countries face acute decisions about how to integrate AI governance into education policy, consumer protection law, and child safeguarding provisions. The governance gaps identified by UNICEF suggest that reactive regulation—responding to harms as they emerge—will be inadequate. Regional governments must instead embed child protection considerations into AI policy architecture from inception, drawing on international expertise while crafting locally relevant standards.
UNICEF's prescriptive recommendations address the institutional ecosystem required to manage these risks. Governments and technology companies must collectively prioritise embedding child rights protection into the design and deployment of AI systems, moving beyond rhetoric toward enforceable standards. Substantial investment in rigorous research examining AI's differential impacts on children of varying ages, abilities, and socioeconomic contexts remains critically underfunded. Legislative frameworks must be strengthened specifically to address AI-enabled sexual exploitation, recognising that existing child protection law often predates these technological capabilities. Education systems require foundational AI literacy curricula that enable young people to comprehend how these systems function and how to evaluate their outputs critically.
The digital divide dimension of UNICEF's analysis deserves particular attention in a region characterised by significant variation in connectivity and technological access. While this report emphasises that children are over-exposed to AI relative to their understanding, the inverse risk exists for children lacking access to beneficial AI applications in education and skill development. Closing this divide requires deliberate investment ensuring that protective frameworks do not become mechanisms for restricting opportunity, but rather create conditions in which all children can engage with AI systems from informed positions.
UNICEF frames the current moment as decisive, emphasising that governance choices made in the immediate term will establish precedents and infrastructures determining children's safety, privacy, and opportunity for decades. This temporal framing underscores that early intervention is substantially cheaper and more effective than remediation after harms have become entrenched. The convergence of technological capabilities, market incentives, and regulatory uncertainty creates a window in which intentional policy design remains possible. Beyond this window lies a scenario where AI governance becomes path-dependent, constrained by incumbent interests and user behaviour patterns difficult to alter retroactively.
