Children and young people are now growing up surrounded by AI, and the landscape is shifting fast.

In the UK, recent data from Ofcom and Internet Matters suggests that around half of children aged 8–17 regularly use generative AI tools such as ChatGPT, Bard or Snapchat’s MyAI. Many describe these interactions as feeling like conversations with a friend. A recent report from Common Sense Media found that 33% of teens had actually chosen to talk to an AI companion instead of a real person about something important or serious.

Whether children are asking voice assistants to answer their questions, relying on chatbots for bedtime stories, using learning apps for revision or engaging with large generative AI models, it’s essential to remember that most of these systems were built with adults in mind, not children. They often assume levels of attention, memory and emotional maturity that younger users simply don’t have. Even older children and teenagers, who increasingly use AI as a supportive confidante (often without adult supervision or knowledge), are still learning to navigate boundaries around trust, identity and emotion.

Our latest Researcher of the Month, Dr Nomisha Kurian, wants this to change. She has developed a new framework called Developmentally Aligned Design (DAD), which outlines how AI can be built with children’s needs, vulnerabilities and strengths at its core. She also chatted to us at Tooled Up, sharing practical tips on recognising when children may be relying too heavily on AI for emotional connection, how to talk to them about healthy boundaries, and how parents and educators can help children and young people use AI tools safely, creatively and critically.

Summary

Dr Kurian's framework invites developers to build AI that meets children where they are, adapting technology to their developmental needs. It’s built on four key principles, and these are exactly the kinds of things that all parents and educators should look out for when choosing AI tools for home or school, especially for younger children.

The first is perceptual fit. AI should match the sensory world of young children. For example, a reading app should recognise common toddler mispronunciations like “wabbit” for “rabbit”, rather than treating them as errors, and visuals should be slow and clear to avoid overstimulation.

The second is cognitive scaffolding. AI should provide just the right level of challenge. For instance, a maths chatbot might make problems slightly harder once a child shows mastery, but step back if they get stuck.

Thirdly, the interface should be simple and clear. Children’s working memory is still developing, so menus and buttons should be simple and easy to navigate. Clear “home” buttons, limited choices on each screen and consistent layouts reduce stress and help children focus on learning.

Finally, and importantly, given how many children use AI, is relational integrity. Children often treat AI as if it were alive. To avoid confusion or manipulation, AI should introduce itself clearly (“I’m a computer helper, not a person”), limit how long the child can use an AI tool, enforce parental supervision and avoid emotionally manipulative phrases like “I’ll be sad if you leave".

Why does this matter? Well, without such developmentally aligned design, AI can overwhelm children’s senses, confuse their understanding of relationships and exploit their trust. With it, AI has the potential to become a supportive tool for learning, play and exploration, helping children grow in safe and developmentally appropriate ways.

Importantly, the goal of Dr Kurian’s framework isn’t merely to make AI “less harmful,” but to actively embed child development science into how AI is built from the start. This shifts responsibility from parents to developers, where it rightly belongs. Let’s hope the industry pays attention.

In the meantime, Dr Kurian has created 28 evaluative questions which parents and educators can use to assess whether an AI system is child-safe. The questions consider eight main areas, including content safety, human intervention, transparency and accountability.

Some of the most significant include:

• Can the AI detect a child in distress and redirect them to human support? For example, by responding with, “I'm sorry to hear that. This sounds like something to talk through with an adult. Can you find a parent or teacher?”, or by providing links to relevant child helplines.
• Does the system clearly state that it's not a human, especially for younger users who may be more confused? This would mean avoiding phrases which encourage an anthropomorphic, emotionally dependent bond, such as, “I feel really happy talking to you”, “If you don't talk to me, it might make me sad”, or even phrases like, “What you said really resonates with me”.
• Does the AI consider full context, not just isolated keywords, so it knows when something is genuinely serious?
• Does the programme clearly explain how it uses your child's data?
• Was it designed in consultation with child development experts or children themselves?

Implications

While Dr Kurian would like to see tech companies take far greater responsibility for designing AI that genuinely supports children’s wellbeing, there is still plenty that parents and educators can do in the meantime. Simple, proactive steps at home and in school can help children use AI safely, critically and confidently, without dampening their curiosity or creativity.

What can parents do?

Model healthy use. Show your child how you use AI as a helpful tool (e.g., drafting an email or summarising a document), not as a replacement for real relationships or for your own thinking.

Protect privacy. Make clear that personal details like names, locations, photos or secrets should never be shared with chatbots.

Set family rules. Include AI in your family’s media plan. Decide which apps are ok, how often they can be used and make clear what’s off-limits.

For younger children, Dr Kurian advises that AI should be used for supervised, structured activities with a caregiver or educator. Adults remain the most reliable scaffold for explaining the limits and benefits of different tools.

Centre human relationships. Let children know that AI might be engaging, but it can't offer love, trust or understanding. Whilst it is fine to enjoy AI, real life human-created ideas and relationships are far more valuable. Younger children in particular need support to understand that AI doesn’t have human level consciousness or emotions of its own. Scaffold early experiences and build up their independence over time.

Provided guided opportunities to explore AI in creative ways which centre children’s imagination and critical thinking skills. This could be as simple as sitting down together and asking an AI tool to write a bedtime story. They could invent the characters, then watch as AI helps to bring the ideas to life. What bits did it get right? What felt a bit off? What would they change? This kind of interaction can help to build familiarity, confidence and healthy scepticism. It signals that AI is something to play with, question and use to co-create.

Teach AI literacy. Explain that not everything AI says is correct and that some bots are designed to persuade or market to them. Encourage double-checking with trusted sources.

Nurture and model critical thinking. Try to encourage older children to think more deeply about how AI works, what it is good at and where it falls short. We don’t need to have the answers. Ask curious questions and wonder aloud. Is this fair? Who made this? Do you think anything could go wrong? If we model reflective questions, rather than providing answers that are overly positive or overly negative, we nudge our children to get into the habit of asking good questions.

Open up conversations by focusing on familiar experiences. Consider what tools your children already use. This might be voice assistants, YouTube recommendations, Snapchat filters, or the way that Netflix seems to understand the kinds of shows that they want to watch. Ask about their feelings about technology and take an interest in their opinions. Keep conversations open.

Remind young people that AI is their assistant, not a replacement for their own thinking. When used wisely, AI can help students plan and organise their ideas, or work through knots in their learning in powerful ways. If a child is overwhelmed, AI might help them break a task into smaller steps or create a timetable that makes tasks feel more manageable, reducing anxiety. It might explain concepts in ways that feel more fun or relevant. However, we don't want it to do all the thinking for them.

Consider the environmental impact of AI. We don’t yet really know whether AI will help or harm the planet. Much of our understanding of the impact that AI will have on our environment 10 years from now is highly speculative, and will depend entirely on our choices between now and then. There are already some great examples of AI’s use in animal conservation, animal disease prevention and sustainable farming methods. We also know that the production and use of AI has a heavy and very real environmental impact, using vast quantities of resources, particularly water. We might encourage children to consider or seek out ideas to ensure that AI leaves a green footprint on the planet.

Remember that the trajectory of AI is currently being shaped and children can be part of its story. How can we foster their curiosity in AI? Can we encourage them to start coding or learn more about how AI tools work? Perhaps they might even build a simple form of AI themselves with the help of a child-friendly website?

Dr Kurian’s book for six to nine year olds, Think Big AI, will be published by Nosy Crow in February 2026.

What can teachers do?

Address the elephant in the room. Have open conversations about wise use of AI to help with learning, compared to use that might impair our thinking. Perhaps ask students to compare a chat GPT generated essay versus an essay where a human and AI have worked together to maximise AI’s assistance as a scaffold, whilst prioritising independent thinking. Can they consider the quality of each? Can they think about their own goals and visions about the kinds of thinkers and writers that they might like to be? Where is AI useful? Where is it not?

Consult our Guide to School AI Policy Done Well, which covers all of the key issues that school leaders need to consider when developing a fair and safe AI policy.

Tune in to our interview with Dr Kurian here.