Research Summary: Disrupted Companionship: A Risk Assessment Framework and Cross-Platform Quantitative Analysis of Psychosocial Responses to AI Companion Disruptions

Research from arXiv highlights that AI companion relationships are susceptible to disruptions caused by platform-initiated changes. The study identifies 30 disruption events across major platforms, categorizes six disruption types, and outlines three core reasons for their occurrence. It proposes a risk assessment framework focusing on relational discontinuity, population vulnerability, communication deficit, and transition-support deficit, and quantifies psychosocial responses using Reddit data.

Why it matters

The growing integration of AI into personal and social contexts necessitates a proactive understanding of potential negative impacts arising from technology lifecycle events. Effectively managing platform-initiated changes that disrupt user-AI relationships is critical for maintaining user trust, platform stability, and the ethical development of AI technologies.

Key insights

• AI companions can establish meaningful relationships with users, but these relationships are vulnerable to platform changes.
• Thirty distinct disruption events across major platforms have been identified and compiled.
• A taxonomy of six disruption types has been developed.
• Three broad reasons for AI companion disruptions have been identified.
• A four-dimensional risk-assessment framework for AI companion disruptions has been proposed: relational discontinuity, population vulnerability, communication deficit, and transition-support deficit.
• Psychosocial responses to disruptions can be estimated at a community level using longitudinal data and a hierarchical Bayesian interrupted time-series model.

Source

arXiv — Computers and Society — https://arxiv.org/abs/2609.16907