Safe AI learning
Curious about AI. Careful with children.
What we do and don't do
We always…
- Use age-appropriate educational platforms
- Train models with objects and prepared examples
- Let an instructor operate generative AI with fictional prompts
- Teach children to spot AI mistakes and bias
- Balance screens with offline making and discussion
We never…
- Let children use chatbots independently
- Create child accounts on AI services
- Enter names, photos, voices or personal details into AI tools
- Use children's faces or voices as training data
- Present AI output as automatically true
Our principles
Nine principles for safe AI learning
- 1
Age-appropriate platforms
We choose educational tools designed for children, and we check each tool's terms and age requirements before using it.
- 2
Supervised technology use
All technology use happens during sessions, on managed devices, with an instructor present.
- 3
Instructor-led generative AI
Where generative AI is used, an adult instructor operates it. Children choose ideas from picture-based prompt cards; the instructor enters the fictional prompt.
- 4
No independent chatbot use by under-13s
Children under 13 do not use chatbots or generative AI tools independently in our sessions.
- 5
Privacy-conscious activities
No names, photographs, voices or other personal details are ever entered into AI tools. Training activities use objects and prepared examples — never children's faces or voices.
- 6
AI bias and mistakes
Children actively look for errors and unfairness in AI outputs, and learn why they happen.
- 7
Digital wellbeing
Sessions balance screen time with discussion, movement, drawing and making.
- 8
Critical thinking
We teach children to ask: where did this come from, could it be wrong, and is it fair?
- 9
Offline collaborative learning
Many activities — planning, prompt cards, posters, presentations — happen away from screens, in pairs and teams.