Safety & Privacy
KAI’™ translates child-protection rules into its design before launch:
how data is collected,
where it is stored, what the AI may see and what parents control.
Child accounts are created by parents only. No child signs up alone: a parent creates the account and gives verified consent, in line with the parental consent requirements of child-protection rules such as COPPA and GDPR-K.
01
Every child account is created by a parent. Access requires verified parental authorization, handled through a specialized verification partner.
02
Age bands instead of full birth dates, pseudonymous child profiles, minimized identifiers and personal data filtered before AI inference.
03
EU child data stays under EU governance and U.S. child data under U.S. governance, with minimized cross-border processing.
04
Only the information each interaction needs reaches the AI. Child data is not used for model training by default.
05
Defined retention rules, parent-controlled deletion, activity visibility and auditable system events.
A Data Protection Impact Assessment has been prepared with the CNIL methodology, covering child-specific AI risks.
Non-negotiable
KAI’™ is a child-first conversational AI designed for pedagogy, safety and healthy interaction. These principles hold regardless of internal implementation choices.
Enforced before and after generation, never bolted on afterward.
Each child operates through a dedicated, supervised profile, with jurisdiction-aware data storage and processing.
Constrained by age band, language, household settings, topic boundaries and risk classification.
Responses come from controlled sources, not unrestricted browsing.
Guiding learning and reasoning through age-adapted interaction.
A core layer of the system, not an optional add-on.
Degraded conditions never increase risk exposure.
Everyday safeguards
Questions about privacy? Write to us at privacy@zenoai.com.