Homework, Advice and Deepfake Fears: The New AI Reality for Children

UNICEF analysis estimates that at least 20 million children across 10 countries have already used artificial intelligence for learning, advice and everyday tasks. The findings point to a widening gap between how quickly children are adopting AI and how slowly governance, safety rules and child-protection systems are adapting to the risks.

Artificial intelligence is already becoming part of children's daily lives, with UNICEF estimating that at least 20 million children across 10 countries have used AI for learning, advice and everyday tasks. Around 13 million children use AI to help with schoolwork and homework, while more than 2 million, or one in every ten, said they turn to AI for advice about problems that worry them.

According to UNICEF estimates, children are adopting the technology faster than adults, making AI governance a child-rights issue rather than only a technology or innovation debate.

AI is already rewiring Childhood

For many children, AI is becoming a routine companion in education and daily life. It can help explain school lessons, generate ideas, answer questions and provide quick support for homework. In some cases, it may give children access to information and learning assistance that would otherwise be unavailable.

However, UNICEF's analysis also shows that AI use is moving beyond classrooms and academic support. When children seek advice from AI about problems that worry them, the technology enters a more sensitive space. The issue is not only whether AI can help with learning, but whether children can safely rely on systems that may influence how they understand information, relationships, risks and personal concerns.

The shift makes child safety a central part of the AI governance debate. Children are not just passive users of technology. They are becoming regular participants in AI-driven environments, often without having meaningful control over how these systems work, how their data is handled or what safeguards are built into the tools they use.

The Promise Comes With a New Risk Layer

AI can create real opportunities for children, especially in education. It can support schoolwork, help explain difficult concepts and offer personalised assistance. For families and schools with limited resources, such tools may appear especially valuable. However, the same technology also brings risks that are harder to manage. UNICEF has warned that children are increasingly exposed to AI systems, their design choices and the collection and use of personal data. Many young users may not fully understand how AI-generated answers are produced, whether the information is reliable, or what happens to the data they provide.

The risks are not abstract. One-third of surveyed children expressed concern about AI being used to create scams or spread misinformation. One in four feared that their images or videos could be manipulated into sexually explicit deepfakes. Deepfakes are synthetic or altered images, videos or audio that can make people appear in content they did not create or consent to.

These concerns show that children are not only using AI; they are also aware that it can be misused. The challenge for policymakers and technology companies is to ensure that this awareness is matched by real protections.

Governance Is Losing Ground to Speed

UNICEF's warning comes as AI adoption among young people moves faster than the systems meant to regulate it. Laws, safety standards and platform protections are still developing, while children are already using AI tools at scale.

Governments must decide how to protect children from AI-enabled exploitation, misinformation, harmful content and privacy risks without limiting access to useful technology. Technology companies must make design choices that account for children's rights, not only user growth or product performance. Parents and teachers must guide children through tools that even many adults are still learning to understand.

Research is another weak point. UNICEF said evidence on AI's long-term effects on children's cognitive development, emotional wellbeing and exposure to harmful content remains at an early stage. That means decisions on AI governance are being made while major questions remain unresolved.

Waiting for perfect evidence could leave children exposed, but weak or poorly designed rules could also fail to address the real risks. The policy task is to build protections that are strong enough to matter, flexible enough to adapt and clear enough for families, schools and companies to follow.

The Child-Rights Test for AI

Ahead of the first Global Dialogue on AI Governance, UNICEF is urging governments, technology companies and international partners to place children's rights at the centre of AI regulation. Its priorities include:

• stronger research into AI's impact on young people
• tougher laws against AI-enabled child sexual exploitation
• safer and more transparent systems
• better digital literacy for children and parents, and
• wider access to reliable digital infrastructure.

AI policy is not only about innovation, competition or productivity. It is also about safety, privacy, well-being and equality. Children who have access to reliable digital tools may benefit from new learning opportunities. Children without access may fall further behind. Children using unsafe systems may face new forms of harm.

The next phase of AI governance will need to answer several urgent questions. How should platforms be designed for younger users? What rules should govern children's data? How can schools use AI without weakening learning or privacy? What protections are needed against deepfakes and AI-enabled exploitation? And how can children's voices be included in decisions about technologies that increasingly shape their lives?