The Addictive Intimacy of AI: Understanding User Disengagement from AI Companions and Why Some Relationships with AI Become Difficult to Leave

Content Warning: This paper presents textual examples that may be offensive or upsetting. AI chatbots are increasingly used as sources of emotional support, on dedicated companion apps and general-purpose assistants alike, yet little is known about what happens when users try to leave. Combining a content analysis of Reddit posts about quitting or reducing use (N=2,782) with interviews with users who found leaving difficult (N=16), we show that disengagement sometimes is not a single decision but a recursive trajectory: triggers prompt users to question the relationship, attempts to leave collide with barriers, and some users cycle through quitting and returning. We propose the notion of the addictive intimacy of AI, a configuration in which the qualities that make a companion emotionally valuable are the same ones that make it harder for users to limit their use and leave, so that intimacy and disengagement risk cannot be treated as independent design problems. We close with design implications for responsible offboarding.

Introduction. Users increasingly interact with conversational AI as sources of emotional support [11, 31, 38]. Such relationships form not only on platforms built for companionship, such as Replika and Character.AI, but also with general-purpose assistants such as ChatGPT and Gemini, where users move fluidly between practical help and emotional support within a single chatbot [6, 24, 32, 38]. The depth of these bonds became more publicly visible when OpenAI retired GPT-4o and users mobilized to demand its return, grieving the loss of a model they had come to regard as a partner [24]. For some users, these relationships provide nonjudgmental support, companionship during loneliness, and a space for self-expression or identity exploration [2, 29], which helps explain why AI companions can feel compelling and difficult to replace. Stepping back, however, is far from straightforward.

Discussion / Conclusion. One pattern explains why some users left without prolonged struggle while others kept coming back: for the deeply engaged, what 5This echoes recent work on LLM refusals in mental health support, which finds that whether a safety response helps or harms depends not on the referral itself but on how it unfolds as an experience, including tiered assessment of actual risk, framing that preserves support rather than terminating it, and resource guidance tailored to the person and moment [50]. The Addictive Intimacy of AI Conference’17, July 2017, Washington, DC, USA they valued in the relationship was also what held them there. It was not the repertoire that separated them. Our interview participants described the same triggers (Section 4) and the same strategies (Section 5) that circulate in the Reddit corpus; what differed was the outcome. Within our interview sample, what separated the attempts that held from those that collapsed was not which strategy users chose, but whether it could survive contact with the relationship. Cognitive Reframing (Section 5.5) shows what surviving took: the accounts in which reframing carried through were ones in which users devalued what the AI was, reinterpreting a partner as “code.”