Safety Approach
Layered protections for children, with practical parent workflows.
Multi-layer policy system
Every child message is evaluated through a policy pipeline that combines baseline safety, family guidance settings, and runtime risk handling. This helps keep conversations supportive and age-appropriate.
Tiered safety events
Parents receive structured safety events instead of a raw surveillance feed. Lower-severity events are summarized. Higher-severity events can show the flagged exchange for context, with a review workflow.
Built for trust, not fear
SapioChat is designed to keep children using healthy AI tools while making it easier for parents to step in when it matters most.
Related pages
A note on AI safety
SapioChat adds guidance and safety layers to AI conversations, but no system is perfect. We use OpenAI's models with family-specific guardrails. AI can occasionally produce inaccurate or unexpected responses despite our best efforts. Parents should maintain open communication with their children about AI interactions and treat SapioChat as one tool among many — not a replacement for family dialogue or professional guidance.