Assumptions and principles for responsible AI development and use in grief and bereavement care

The rapid integration of artificial intelligence (AI) into grief and bereavement care represents a significant technological shift with therapeutic potential and ethical complexity. This paper establishes evidence-informed assumptions and principles to guide the responsible development and use of AI technologies in grief support, addressing the needs of bereaved individuals, families, communities, professionals, technology developers, and policymakers. We define "Grief AI" as applications spanning five core technological domains: Natural Language Processing for information seeking and therapeutic conversations; Computer Vision for emotion analysis and immersive experiences; Machine Learning for trajectory forecasting and treatment matching; Conversational and Generative AI for grief bots; and Integrated Multimodal AI incorporating advanced autonomous systems. The paper classifies these applications into low-, medium-, and high-risk categories based on factors including depth of user engagement, AI autonomy, potential for psychological harm, level of human supervision, and posthumous consent considerations. The paper acknowledges unresolved tensions regarding the authenticity of AI-generated empathy, cultural evolution versus inappropriate technologization of sacred experiences, implications for healthy grief adaptation, environmental ethics, and professional transformation. These provocations highlight the need for ongoing dialogue as the field evolves. We conclude that AI technologies can complement human-centered grief support through responsible development characterized by interdisciplinary collaboration, transparency, cultural sensitivity, and an unwavering commitment to human dignity and welfare over commercial interests.