Why children’s perspectives belong at the centre of AI policy and design

Children are actively using Artificial Intelligence (AI), and the world is yet to catch up with what this means for their rights, development and well-being. 

As AI becomes more embedded in children’s lives, their voices remain largely absent from the policy debates shaping its development and use. Now more than ever, policies governing digital technology and their impact on children’s rights must be adaptable, forward looking and based on children's lived experiences. These policies should be updated regularly to remain relevant and effective in protecting children’s rights amidst ongoing technological advancements.

Whether AI is helping or harming children, we turned to children to help us understand if their interests and well-being are considered by the digital technologies they’re actively using. Looking back to child consultations held in 2020 and 2025, we identified several notable changes in how children use AI and how they think about it. Five key themes and shifts stood out:   

Children remain excluded from AI policy and system development

In 2020, children stated strongly that decision-making about AI is adult-centric. In the 2025 consultations, children raised the same concern. The UNICEF 2024 Global Outlook observed that “despite the advances, attention to children’s rights and needs in national AI strategies and policies is still limited”. 

A report from Kenya last year by the Digital Futures for Children highlights this systemic problem: “There is a huge gap between the best interests of the child and having the views of the children heard. Children are not only the majority of users of AI tools, they are also the faster adopters, yet the current ecosystem does not include them in the design, development and regulation of AI.”

Child participation means giving children the agency and opportunity to shape the policies and decisions that affect them directly in accordance with their evolving capacities. UNICEF’s Guidance on AI and Children provides recommendations for how to make such engagement meaningful. To break the adult-centricity in AI, children also need to know their rights so that they can demand to be included in decisions-making and be empowered to defend their rights when needed.

AI no longer lurks in the background, it is front and centre

In 2020, AI was largely in the background and children encountered it via recommendation algorithms. Five years later, in 2025, children were interacting directly with AI in a range of ways – such as for homework, creativity, advice, emotional support and even friendship. 

AI is no longer behind the scenes. It’s a tool children interact with daily. 

For some children, AI has become a companion and brings a risk of emotional dependence. In Europe, around seven in ten children report using some form of generative AI, either as a standalone chatbot or embedded directly within the platforms children use, according to research by the London School of Economics.

AI education in schools is growing, but gaps remain in practice and in policy

In the 2020 consultations, it was apparent that most children had learned about AI in their homes and among peers, rather than in school. This has started to change, and AI is slowly becoming part of the education curricula in some countries. 

The design and pace of AI adoption in national education systems vary across countries: China established a tiered AI education system starting from primary school in 2025. India announced that it will introduce “Artificial Intelligence and Computer Thinking” in schools starting from Grade 3. In the United Arab Emirates, AI is now a required subject in every public school from kindergarten to Grade 12. 

These are some examples of AI integration into the education system. Many countries are still catching up, as AI education remains insufficiently integrated into national curricula. These uneven approaches to AI education could result in an AI divide between children who know how to effectively, safely and ethically use the technology, and those who don’t. 

Beyond the classroom, these tensions are also evident in the policy frameworks shaping children’s role in an AI-enabled future. In 2020, many national AI strategies focused on children as a future workforce. In 2025, this framing of children as an economic resource persisted, with digital inclusion policies also casting children as victims in need of protection.

Although education is a crucial foundation for helping children develop digital skills that enable them to navigate their lives, focusing too narrowly on preparing them as workers or only protecting them limits the full range of rights children have to fully experience childhood. 

Changing views about AI

In the 2020 child consultations, participants expressed concern about AI-driven automation leading to job losses and a few had critical understandings and concerns on AI. 

But this too has shifted, and in 2025 the conversations involved AI-generated misinformation and hallucinations. Children talked about the limitations of AI in content filtering and age estimation, and worried about how using AI can undermine learning and critical thinking. Children were also concerned about harmful content generated with AI – such as online child sexual exploitation and abuse content, and deceptive deepfakes.

The 2025 consultations also showed a much deeper emotional integration of AI, as some children turn to AI for advice and emotional support. UNICEF’s research in ten countries shows one in ten AI-using children turn to AI for advice on things that worry them. Upholding children’s best interests must now account for social-emotional development vis-à-vis AI in a way that was not anticipated back in 2020.

Knowledge of risks and shared responsibility

In 2020, most children believed that parents had the primary responsibility for their protection in an AI world, and in 2025, for some children, this vision has shifted towards a shared responsibility, which includes parents, children, technology companies, schools, and governments.

In the 2025 consultations with youth, “participants envisioned a future where responsibility for the best interests of children in the digital environment was shared across generations, governments, platforms, and communities.”

Shared responsibility implies children as key stakeholders in decision-making on all levels. This resonates with what children expressed in 2020: Some believed they had agency to confront AI risks and opportunities as individuals. Others believed that they could act collectively to tackle risks posed by AI.

Why must we listen to children and young people?

While a generation grows up inside a global experiment, digital technologies continue to develop faster than regulation and enforcement can keep pace. To uphold children’s rights in such a landscape, best interests prioritizations must be iterative and continuous, not a ‘one-and-done’ checklist. That is why it's important to engage children in foresight exercises to get their views on the digital future that would uphold their rights.

As it currently stands, the ‘best interests of the child’ principle mostly comes up in relation to the present technologies in policy and regulation. In this fast-paced digital landscape, policies must address the impact of AI on children’s rights and ensure that they are resilient, adaptable and future-proof. 

For more information on children’s best interests, read General Comment No 14 and UNICEF’s working paper on this topic. This article is a reflective review of evidence from two different child consultations that took place in 2020 and in 2025. Findings from a similar engagement of children and youth and AI in 2025 also informed this article. Learn more about UNICEF Innocenti’s Best Interests of the Child in the Digital Environment project.