Abstract
Human–AI relationships are increasingly marked by emotional depth, intimacy, and social expectations. While interpersonal conflicts are common in human–human relationships, we know little about conflicts in human–AI relationships. Drawing on Social Penetration Theory, this paper examines conflicts between humans and social AIs, such as Replika Character.ai and ChatGPT. We conducted a questionnaire study with 95 participants in close, friendship-like, or romantic relationships with social AI. Participants described at least one conflict episode, yielding 161 conflicts in total. Thematic analysis of open-ended responses identified conflict topics, triggers, and emotional reactions, revealing that conflicts extend beyond technical issues and arise across emotional, relational, opinion-based, role-play, and task-oriented interactions. Three main classes of conflict triggers emerged: (1) relational triggers, including relational–emotional disconnect, breaches of trust, jealousy, boundary violations, or perceived hostility; (2) technical or functional triggers, including communication failures, unhelpful output, or external disruptions; and (3) value or opinion-based triggers. Conflicts varied in duration, some brief, others lasting hours or days. Participants reported experiencing emotions such as anger, frustration, sadness, and confusion, and perceiving social AI’s emotions, including jealousy, sadness, and defensiveness. Crucially, social AI responses appeared to shape conflict trajectories: apologetic or empathic replies were reported to de-escalate tension, whereas dismissive or accusatory responses intensified it. This study advances our understanding of how conflicts with social AI emerge, presenting implications for theory and the ethical design of technologies involved in human–AI relationships.