Solutions stage
Panel
In person
LeadersGoldDiscovery

Toward beneficial AI for children, Defining and scaling model behaviors assessment

  • Date
    9 July 2026
    Timeframe
    14:30 - 15:00 CEST
    Duration
    30 minutes

      AI has the potential to expand children’s access to learning, creativity, and support, but only if it is designed around their developmental needs. As systems become more conversational and relational, child AI safety cannot stop at filtering harmful content. We also need to evaluate how models behave: whether they support autonomy, mastery, and real-world connection, or whether they create false intimacy, dependency, and inappropriate parasocial pull.

      In this session, Gregory Renard and Mathilde Cerioli will introduce their vision for promoting beneficial AI for children, through a behavioral assessment methodology that translates developmental science into operationalized measures, that support product teams, researchers and policymakers.

      The goal is to move from safe AI to AI that is genuinely beneficial for children.

      Are you sure you want to remove this speaker?