The rapid integration of artificial intelligence into children's daily lives has far outpaced adult adoption rates, with young users embracing the technology at a velocity more than three times greater than their older counterparts. This striking disparity emerged from fresh data compiled by the United Nations Children's Fund ahead of the inaugural Global Dialogue on AI Governance, raising urgent questions about digital protection and childhood safety in an increasingly automated world.
UNICEF's findings, drawn from research across 10 countries, paint a picture of a generation already embedded within AI ecosystems without corresponding safeguards. At least 20 million children globally have engaged with AI technologies, a substantial figure that underscores how comprehensively these tools have penetrated young people's digital environments. For many, AI has become an accessible source of support and guidance—over two million children, representing roughly one in ten across the surveyed populations, report turning to AI systems for counsel on matters causing them anxiety or concern.
The application of AI in education represents another significant dimension of childhood adoption. Approximately 13 million children across the surveyed nations utilise artificial intelligence to assist with learning activities and homework completion. This educational dimension reflects how seamlessly AI has woven itself into formal and informal learning structures, offering students instant access to information and problem-solving assistance that previous generations lacked. Yet this accessibility, while ostensibly beneficial for academic outcomes, introduces novel vulnerabilities that educational institutions and parents remain largely unprepared to address.
A fundamental tension underlies this expansion: children encounter AI systems with minimal understanding of their mechanics, financial incentives, or data practices. They possess limited capacity to opt out of these digital ecosystems or meaningfully challenge the systems shaping their online experiences. This power imbalance places young users in particularly precarious positions, as they navigate platforms and tools designed primarily for profit extraction rather than their welfare. UNICEF emphasises that this structural disadvantage means children absorb the consequences of inadequate AI governance most acutely and will endure these repercussions throughout their lifetimes.
The security risks associated with unregulated AI use among children are substantial and multifaceted. One-third of young people surveyed across the participating countries expressed worry that AI could be weaponised for deception—either through manipulative scams targeting minors directly or through the amplification of false narratives that corrode their understanding of reality. Such concerns are not speculative; deepfake technology and content fabrication capabilities have already demonstrated capacity to harm vulnerable populations. Even more alarming, approximately one-quarter of surveyed children reported anxiety about AI-generated sexual abuse material—a category of content created by manipulating genuine images or videos of real people into explicit scenarios. This threat carries devastating psychological and legal consequences, yet remains inadequately addressed in most jurisdictions' AI regulatory frameworks.
The architecture of current AI deployment reveals what UNICEF characterises as a troubling absence of protective mechanisms. Safety considerations appear peripheral rather than central to system design, with companies and developers releasing increasingly powerful tools into public circulation despite incomplete understanding of downstream consequences for minors. This gap between technological capability and protective infrastructure creates what amounts to an experiment conducted on children without their informed consent or meaningful parental oversight. The stakes of this negligence extend beyond individual harm to encompass broader social consequences, including erosion of information reliability and normalisation of consent violations.
Addressing these challenges requires coordinated action across multiple sectors and jurisdictional boundaries. UNICEF has outlined a comprehensive agenda encompassing government regulation, corporate responsibility, and international cooperation. Substantial investment in research specifically examining AI's risks to children remains inadequate; policymakers currently operate with incomplete evidence about developmental impacts and vulnerability factors. Simultaneously, existing laws prohibiting child sexual exploitation require modernisation to encompass AI-generated and AI-manipulated content, closing legal ambiguities that currently shield perpetrators.
Transparency and safety must become non-negotiable features of AI systems accessible to children, rather than optional enhancements introduced after public pressure. Developers should be compelled to conduct child-impact assessments prior to deployment and maintain ongoing monitoring of actual usage patterns and harms. Building digital literacy among young people represents another critical intervention, enabling them to recognise manipulation, understand data extraction practices, and make more informed choices about technology engagement. Yet literacy initiatives prove insufficient without simultaneous efforts to reduce technological access barriers—the digital divide continues widening, with economically disadvantaged children in developing nations potentially excluded from educational and economic opportunities created through AI, even as they remain exposed to associated risks.
For Malaysian policymakers and regional stakeholders, UNICEF's findings carry particular resonance given the rapid proliferation of digital platforms across Southeast Asia and varying maturity levels of protective legislation. The Malaysian government and neighbouring administrations must urgently examine whether existing child protection frameworks adequately address AI-specific harms, or whether new legislation tailored to this technology's unique characteristics is required. Industry self-regulation has repeatedly proven inadequate in protecting vulnerable populations; meaningful governmental intervention supported by adequate enforcement resources will likely prove necessary.
The temporal dimension of this challenge cannot be overstated. AI governance decisions made in coming months will establish precedents and technological trajectories affecting children's safety, privacy, and opportunity access for decades. This window for shaping AI development toward child-protective outcomes remains open but is rapidly closing as entrenched interests and technological lock-in accumulate. UNICEF's warning amounts to a call for immediate, decisive action rather than gradual adaptation—a recognition that the scale and speed of AI's integration into childhood demands response proportionate to the stakes involved.
