Policymakers, academics, healthcare providers, AI developers, and patient advocates convened by Stanford HAI identify critical gaps in how we regulate AI tools used for therapy and emotional support.