A guide for the grown-up beside them.
A little preparation helps make room for curiosity, careful checking, and real conversation.
Small activities. An adult who stays involved.
Start with one video and its matching game. Ask the child what they noticed; invite them to explain an answer rather than perform for a score. Take breaks and adapt the reading to the learner.
- Ages 5–7: read aloud, point to a shape, and use familiar make-believe examples.
- Ages 8–10: stay alongside the learner; practice invented details and checking a short source note.
- Ages 11–12: explore together; ask what is missing and which source supports a claim.
- Ages 13–17: discuss tool age requirements, school expectations, attribution, privacy, and when an adult should help.
These age paths are editorial teaching suggestions, not legal eligibility rules or a claim that all children at one age need the same support.
The live demonstration is archived.
Our live-model service is offline. The saved games and prepared examples still work without a live model.
What these games do with information
The three games have fixed choices and programmed feedback. They do not send answers to an LLM, ask for names or birthdays, use accounts, keep scores, store game answers between visits, or include advertising or analytics code. Game state stays in the open page. Reset or closing the page clears it.
Loading pages and video files still makes ordinary requests to our website; hosting systems may keep technical logs. This is not a claim of anonymous browsing or zero data processing. Read the site privacy notice for information about ordinary website requests and the archived live tools.
Keep people, not the chatbot, responsible.
UNICEF’s guidance places children’s safety, privacy, and best interests at the centre. eSafety warns that many generative tools lack safeguards designed for children and that controls are only one part of protection. Adult attention remains necessary.
UNESCO advocates human-led, age-appropriate educational use; its guidance is not a universal legal age rule. Check the service’s current requirements and your school’s rules. eSafety’s media-literacy advice supports checking dates and original evidence, and seeking a trusted adult when something is upsetting.
Pause if a child becomes distressed, treats a chatbot as a secret confidant, or feels pushed to share personal information. Offer real-world support. Do not use this corner or a live model for urgent or professional decisions.
Explore the wider AI safety hub →Sources for grown-ups
Checked October 16, 2026. These sources inform the guidance; their authors have not endorsed this website.
- UNICEF: Guidance on AI and children, version 3. December 2025.
- eSafety: Parental controls, including generative AI. Updated September 10, 2026.
- UNESCO: Guidance for generative AI in education and research. 2023 guidance; page updated January 16, 2026.
- eSafety: Critical thinking and deciding what to trust. Updated October 16, 2026.