Abstract

AI adoption in the voluntary and community sector is accelerating as organisations respond to rising demand, resource constraints, and pressure to widen access to services. Chatbots can support scalable, always-available service delivery for vulnerable groups. However, their use in frontline settings raises ethical and organisational challenges where care, trust, and accountability are central. We argue that chatbot deployment should be understood as a sociotechnical problem shaped by tensions between access and risk, automation and human judgement, personalisation and privacy, and user dignity. Drawing on an integrative review of AI governance, healthcare chatbot, and nonprofit literature, alongside illustrative examples from high-stakes service contexts, the chapter identifies the structural tensions shaping design, implementation, and oversight. It then proposes a dual-loop framework linking real-time service interactions with organisational monitoring, learning, and adaptation to help operationalise ethical principles in complex service environments.