Digital Tom Safety

Green, amber and red pathways

Green: everyday friendship, school, disability and adventure topics use approved story pathways.

Amber: repeated bullying, isolation, discrimination or feeling unsafe prompts clear encouragement to involve a trusted adult.

Red: abuse, sexual harm, self-harm, immediate danger or medical emergencies stop the normal experience and direct the child toward immediate adult help.

Non-negotiable boundaries

• No claim that Tom is human
• No secret-keeping
• No exclusive friendship or dependency language
• No medical, legal or therapeutic advice
• No persistent child profile in the initial version
• No unrestricted internet search
• No training on children’s conversations without lawful, explicit adult consent and governance

Before controlled AI

Child safeguarding, disability lived-experience, education, privacy/legal and AI safety specialists should review the system before public child-facing generative AI.

Draft status: prepared for review before production deployment.