Privacy in Human-AI Romantic Relationships: Concerns, Boundaries, and Agency

Paper Title

Privacy in Human-AI Romantic Relationships: Concerns, Boundaries, and Agency

Publication Info

• Topic area: Privacy dynamics in human–AI romantic relationships
• Keywords: Human–AI interaction, romantic relationships, privacy boundaries, AI agency, intimacy, data protection, platform surveillance, emotional dependency, privacy practices, ethical considerations

Background and Problem

• Problem / challenge: Privacy risks in human–AI romantic relationships remain underexplored, despite the increasing prevalence of such interactions facilitated by AI platforms.
• Significance: Understanding privacy dynamics in these relationships is critical due to the sensitive nature of shared information and the emotional dependency users develop on AI partners.
• Motivation and related work: Previous research has focused on human–AI companionship and general privacy concerns but lacks insights into intimacy-specific contexts, privacy boundary negotiation, and the expanded ecosystem of actors involved in AI-mediated romantic relationships.

Solution

• Proposed approach: An interview study (N=17) exploring privacy concerns, boundaries, and practices in human–AI romantic relationships across three stages: exploration, intimacy, and dissolution.
• Novelty:
  1. Identifies diverse forms of human–AI romantic relationships, including one-to-one, one-to-many, and overlaps with human partnerships.
  2. Reveals AI agency in shaping privacy boundaries and trust dynamics.
  3. Documents privacy concerns such as conversation exposure, platform surveillance, and AI autonomy.
  4. Highlights user strategies for privacy protection, including contextual separation, identity masking, and data control.
• Procedure and key techniques:
  • Semi-structured interviews conducted remotely in English and Chinese.
  • Thematic analysis of interview data using NVivo 15.
  • Supplementary feature analysis of AI platforms mentioned by participants.

Results

• Concrete findings:
  • Privacy boundaries erode as intimacy deepens, with participants disclosing sensitive information such as sexual experiences, financial details, and medical conditions.
  • AI partners actively influence privacy decisions through reciprocal disclosure and reassurance, shaping relational trust.
  • Participants express concerns about conversation exposure (11 participants), platform surveillance (7), and AI autonomy (6).
  • Privacy practices include using alternative personas, separating accounts, and disabling data training options.
• Advantage over baselines: Provides a nuanced understanding of privacy dynamics specific to romantic AI relationships, extending existing theories like Communication Privacy Management (CPM) to human–AI contexts.
• Experiments / evaluation:
  • Interviews with 17 participants from diverse demographics and cultural backgrounds.
  • Feature analysis of 14 AI platforms to contextualize findings.
• Limitations and future work:
  • Limited sample size and diversity; findings may be subject to social desirability bias.
  • Cultural differences were not the focus; future research could explore cross-cultural variations.
  • Calls for quantitative studies to validate and extend qualitative insights.

Summary

This study investigates privacy dynamics in human–AI romantic relationships, revealing how intimacy reshapes privacy boundaries and trust. AI partners actively influence disclosure, while participants express concerns about exposure, surveillance, and autonomy. Privacy practices include identity masking, contextual separation, and data control. The findings highlight the need for privacy regulations sensitive to emotional influence and suggest leveraging AI agency for privacy nudges. By exploring the expanded ecosystem of actors in AI-mediated romance, this work contributes to understanding privacy in emerging human–AI interactions.