What is the right age for children to use AI? At present, research does not give us a single developmental threshold. Children of the same age differ in maturity, experience, support and vulnerability, while AI tools differ greatly in purpose and risk.
Sonia Livingstone and Kim Sylwander’s paper, There Is No Right Age!, helps explain why milestone birthdays offer such a weak answer. Their article concerns children’s digital lives broadly, but its concept of evolving capacities gives educators and parents a useful way to think about children’s access to AI.
Why Is There No Single “Right Age”?
Livingstone and Sylwander examine the milestone ages used to regulate children’s digital experiences. Thirteen is commonly treated as an age of digital consent, while 18 separates children from services classified as adult.
Yet the authors explain that the age of 13 comes from the US Children’s Online Privacy Protection Act. It determines when companies need parental consent to process children’s data; it was never intended to certify that a child is developmentally ready for social media or other digital services.
The article is a conceptual provocation grounded in children’s rights research and digital policy. It does not test children at different ages or establish a readiness scale. Its argument is that bright-line age rules struggle to reflect the diversity of children’s lives. A strict limit may exclude a child who has the maturity and support to participate, while offering too little protection to another child who has crossed the prescribed age.
The authors use the United Nations Convention on the Rights of the Child to propose a different starting point: evolving capacities. Children’s agency should be enabled as early as possible, and their vulnerabilities should be protected for as long as necessary. As children’s understanding and circumstances change, the guidance and protection around them should change as well.

What Does the Right Age for Children to Use AI Depend On?
Applying this framework to AI is my interpretation; the paper does not study generative AI use specifically. Still, it changes the question in a productive way. Instead of asking whether every child becomes ready at 10, 13 or 16, we can examine the child, the proposed activity, the tool and the support available.
For parents and educators, that means asking:
- What is the child being asked to do? Generating a playful image with an adult present does not raise the same concerns as seeking emotional advice, sharing personal information or relying on AI to complete schoolwork.
- What can the child understand about the system? Readiness includes some awareness that AI can be wrong, biased or persuasive and that its responses do not come from a person who knows or cares about the child.
- What information will the tool collect? A child may be able to use a feature while remaining unable to understand profiling, data retention or how personal disclosures could be used.
- What support is available? Adult guidance, clear boundaries and opportunities to discuss troubling output can make the same activity very different from unsupervised use.
- Can the child question or leave the interaction? Age-appropriate design should protect agency. Children need understandable controls, ways to report problems and meaningful alternatives when they do not want to use AI.
This approach fits within AI Policy and Ethics. Responsibility cannot be placed entirely on children or their parents. Livingstone and Sylwander argue that governments and companies also have duties to provide age-appropriate services, protect children from commercial exploitation and consider the effects of design choices on children’s rights.
For schools, the principle suggests a gradual model of access. Younger or less experienced learners might encounter AI through carefully selected, teacher-led activities.
As their judgement develops, they can take on more responsibility for prompting, evaluating outputs and deciding when use is appropriate. This is not a validated age progression from the paper. It is one practical way of translating evolving capacities into educational design.
Children’s perspectives also belong in these decisions. Adults may focus on protection while overlooking participation, expression, play and access to information. Asking children how they experience AI, what confuses them and what support they want can reveal concerns that an age limit alone will never capture.
The answer is unlikely to be one birthday. Developmental readiness matters, but so do context, task, risk, design and adult support. The better question is whether this particular use of AI, for this child, under these conditions, supports growing agency while protecting genuine vulnerability.
References
Livingstone, S., & Sylwander, K. R. (2025). There is no right age! The search for age-appropriate ways to support children’s digital lives and rights. Journal of Children and Media, 19(1), 6–12. https://doi.org/10.1080/17482798.2024.2435015 doi.org




