Exploring the Interaction of Explanation Styles, Context, and Trust of AI Privacy Redaction in AI-mediated Interactions

2026-07-07

AI Summary

This study investigates how generating interpretable privacy redaction behaviors in AI-mediated communication can enhance user trust while balancing transparency and privacy protection. We design a verifiable system that integrates privacy redaction techniques with a natural language generation module to provide multi-granularity explanations of redacted content. Through a user study involving 180 participants, we evaluate the system’s effectiveness and find that such explanations significantly improve users’ perceived privacy protection (p<0.05, d≈0.3), particularly in high-redaction scenarios where they are perceived as more helpful (p<0.05, f≈0.2). The results further reveal that contextual and individual factors—such as redaction intensity, user age, and familiarity with AI—moderate the efficacy of explanations, suggesting that adaptive, context-aware explanation mechanisms are essential for optimizing trustworthy AI design.

Abstract

AI-mediated communication is increasingly being utilized to help facilitate interactions; however, in privacy sensitive domains, an AI mediator has the additional challenge of considering how to preserve privacy. In these contexts, a mediator may redact or withhold information, raising questions about how users perceive these interventions and whether explanations of system behavior can improve trust. In this work, we investigate how explanations of redaction operations can affect user trust in AI-mediated communication. We devise a scenario where a validated system removes sensitive content from messages and generates explanations of varying detail to communicate its decisions to recipients. We then conduct a user study with 180 participants that studies how user trust and preferences vary for cases with different amounts of redacted content and different levels of explanation detail. Our results show that participants believed our system was more effective at preserving privacy when explanations were provided (p<0.05, Cohen's d ~ 0.3). We also found that contextual factors had an impact; participants relied more on explanations and found them more helpful when the system performed extensive redactions (p<0.05, Cohen's f ~ 0.2). We also found that explanation preferences depended on individual differences as well, and factors such as age and baseline familiarity with AI affected user trust in our system. These findings highlight the importance and challenge of balancing transparency and privacy in AI-mediated communications and suggest that adaptive, context-aware explanations are essential for designing privacy-aware, trustworthy AI systems.

Problem

• AI-mediated communication
• privacy redaction
• explanation styles
• user trust
• context-aware explanations

Innovation

• AI-mediated communication
• privacy redaction
• explanation styles
• user trust
• context-aware explanations