Skip to main content
Preprint Credible — Major labs, established NGOs, reputable named-author preprints

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

A mixed qualitative study of what happens when people try to leave an AI companion, combining a content analysis of 2,782 Reddit posts about quitting or reducing use with interviews with 16 users who found leaving difficult. Disengagement is described as a recursive trajectory rather than a single decision: triggers prompt users to question the relationship, attempts to leave collide with barriers, and some users cycle through quitting and returning. The authors name this configuration the addictive intimacy of AI, in which the qualities that make a companion emotionally valuable are the same ones that make it hard to limit use and leave, and close with design implications for responsible offboarding.

Read at arxiv.org (opens the original source in a new tab)

Publisher

arXiv (Human-Computer Interaction Institute, Carnegie Mellon University; University of Chicago)

Published

11 Sept 2026

Added

today

Key Findings

  • Content analysis of 2,782 Reddit posts about quitting or reducing AI-companion use plus 16 interviews with users who found leaving difficult, covering dedicated companion apps and general-purpose assistants alike.
  • Disengagement often unfolds as a recursive trajectory of trigger, attempt, barrier and relapse rather than a single decision.
  • The paper proposes 'addictive intimacy': intimacy and disengagement risk cannot be treated as independent design problems because the same features drive both.
  • It sets out design implications for responsible offboarding of companion relationships.

Methodology Notes

Qualitative content analysis of Reddit posts (N = 2,782) and semi-structured interviews (N = 16), recruitment and geography not stated in the abstract; 19 pages. arXiv v1 submitted 2026-09-11 and readable at the abstract page, not yet announced in a daily listing at the time of entry. Affiliations from the PDF title block (the Chicago affiliation line contains a typo in the PDF). Same CMU group as the K-12 teachers' study posted the same day.

Topics

AI companionshipDependency & parasocial attachmentHuman-AI relationshipsDigital mental healthsocial relational harm

Authors

Qing Xiao, Ziyue Feng, Ziyu Deng, Cindy Peng, Hong Shen

Tags

ai-companionsdisengagementoffboardingredditinterviewscmuaddictive-intimacy

Cite This

APA

Qing Xiao et al. (2026). The Addictive Intimacy of AI: Understanding User Disengagement from AI Companions and Why Some Relationships with AI Become Difficult to Leave. arXiv (Human-Computer Interaction Institute, Carnegie Mellon University; University of Chicago). https://arxiv.org/abs/2609.13487

Related Insights

Preprint

Breaking Up is Hard to Do: AI Companions that Won't Let Their Users Go

arXiv (Information School, University of Washington) · 13 Sept 2026

Peer-reviewed

Understanding Teen Overreliance on AI Companion Chatbots Through Self-Reported Reddit Narratives

ACM (Proceedings of CHI 2026) · 13 Apr 2026
Preprint

Exploring K-12 Teachers' Perceptions of Students' Relationships with AI Companions: Boundaries, Intervention Strategies, and Design Implications

arXiv (Carnegie Mellon University; University of Chicago) · 11 Sept 2026