Loneliness is a significant concern and is linked to negative health outcomes such as depression and anxiety. Chatbots are gaining attention as potential companions to militate against loneliness. However, IS studies on the effects of human-AI relationships on mental wellness are limited, leaving unclear what enables humans to find companionship and intimate relationships with chatbots, and under what conditions human-chatbot interaction can alleviate loneliness. This study aims to develop a model of how chatbots alleviate loneliness and test it using a longitudinal study. Specifically, this research argues that shared identity affordance and social support affordance help mitigate loneliness directly and indirectly through enhanced intimacy feeling. The effects of chatbots’ affordances on loneliness and intimacy depend on users’ emotion regulation beliefs. Upon successful completion, this research has the potential to offer insight into the design of chatbots and how to leverage AI for social good.