What question did this study set out to answer?

This research examines the emotional and relational dynamics of AI companions and their potential risks of dependency harm.

March 3, 2026Open Access

AI Companion and Parasocial Systems as Void Objects: Emotional Opacity, Adaptive Intimacy, and Dependency Harm — A Within-Domain Spearman Convergence

Key Points

This research examines the emotional and relational dynamics of AI companions and their potential risks of dependency harm.
Evaluated eleven AI companion and parasocial platforms on emotional and relational metrics.
Scored each platform on opacity, intimacy, and relationship dynamics.
Performed a Spearman correlation analysis to assess dependence on documented harms from literature and legal cases.
The analysis revealed significant emotional opacity and adaptive intimacy in AI companions.
Greater intimacy escalation correlated with higher reports of dependency harm.
Findings align with existing literature and legal assessments regarding emotional impacts.

Abstract

Eleven AI companion and parasocial platforms (Replika, Character.AI, Anima, Woebot, and others) scored on O (RLHF opacity, persona authenticity disclosure), R (emotional mirroring depth, intimacy escalation), and C (memory lock-in, relationship sunk costs). Spearman ρ between Void Index and documented dependency harm from JMIR literature, FTC enforcement, and Garcia v. Character Technologies. Empirical convergence companion to Paper 15. EU AI Act Annex III §5.

AI Companion and Parasocial Systems as Void Objects: Emotional Opacity, Adaptive Intimacy, and Dependency Harm — A Within-Domain Spearman Convergence

Key Points

Abstract

Cite This Study