The Impact of Identity Disclosure on Unethical Behavior in AI-Powered Online Customer Service

다국어 초록 (Multilingual Abstract)

The rapid development of artificial intelligence (AI) technology is profoundly reshaping the service model of online customer service, with consumers increasingly interacting with AI-driven virtual customer service agents on e-commerce platforms. Against this backdrop, whether businesses should explicitly disclose the AI nature of their customer service agents to consumers has become a critical issue that concerns both transparency and trust while also implying ethical risks. Existing research has largely focused on the positive effects of identity disclosure on user trust and acceptance, but has paid less attention to the potential for moral laxity and tendencies toward unethical behavior that it may trigger. Based on social cognitive theory and moral disengagement theory, this study constructs a moderated mediation model of “identity disclosure → moral disengagement → unethical behavior” to systematically explore the mechanisms through which the disclosure of AI customer service agents’ identities influences consumers’ unethical behavior, and to examine the moderating role of self- control in this context. This study employed a single-factor, two-level (AI identity disclosure vs. no AI identity disclosure) between-subjects experimental design. Using a simulated e-commerce return scenario as the stimulus context, we collected valid questionnaires from 120 Chinese consumers via an online experiment. Data analysis and hypothesis testing were conducted using SPSS 27.0 and the PROCESS macro. The experimental results indicate that: (1) AI identity disclosure significantly increased consumers’ propensity for immoral behavior; the immoral behavior score in the AI disclosure group (M = 4.71) was significantly higher than that in the non-disclosure group (M = 3.93), with an effect size reaching the large effect level (Cohen’s d = 1.15); (2) Moral disengagement partially mediates the relationship between identity disclosure and immoral behavior, with an indirect effect of -0.305 and a 95% Bootstrap confidence interval of [-0.550, -0.089], which does not include zero; (3) Self-control significantly negatively moderated the relationship between moral disengagement and immoral behavior (interaction term β = -0.139, p = 0.009). The effect of moral disengagement was stronger among individuals with low self-control (β = 0.582, p < 0.001), while the effect among those with high self- control remained significant but was markedly weakened (β = 0.234, p = 0.049). Theoretically, this study expands the ethical dimension of identity disclosure research by introducing moral disengagement theory into the new context of AI-human interaction and deepens our understanding of self-control as a boundary condition. Practically, the findings provide empirical evidence for companies to develop contextualized AI identity disclosure strategies, embed ethical reminder mechanisms in customer service interfaces, optimize anthropomorphic design, and implement differentiated interventions based on user profiles. This study reveals the “ethical paradox” of AI transparency—namely, that while information transparency fosters trust, it may also weaken moral constraints—and calls for responsible AI system design to seek a dynamic balance between technical transparency and psychological ethics.

목차 (Table of Contents)

• ABSTRACT 4
• 1. INTRODUCTION 8
• 1.1 RESEARCH BACKGROUND 8
• 1.2 RESEARCH OBJECTIVES AND SIGNIFICANCE 16
• 2. RELEVANT THEORIES AND LITERATURE REVIEW 29
• 2.1 IDENTITY DISCLOSURE 30
• 2.2 MORAL DETACHMENT 37
• 2.3 CONSUMER UNETHICAL BEHAVIOR 46
• 2.4 SELF-CONTROL 50
• 3. HYPOTHESIS DEVELOPMENT 57
• 3.1 IDENTITY DISCLOSURE AND UNETHICAL CONDUCT 57
• 3.2 THE MEDIATING ROLE OF MORAL DISENGAGEMENT 60
• 3.3 THE REGULATORY ROLE OF SELF-CONTROL 63
• 4. DEVELOPMENT OF STIMULANTS 66
• 5. EXPERIMENT 70
• 5.1 OBJECTIVES OF THE EXPERIMENT 70
• 5.2 EXPERIMENTAL DESIGN AND SAMPLES 71
• 6. ANALYSIS OF EXPERIMENTAL RESULTS 75
• 6.1 RELIABILITY AND VALIDITY ANALYSIS 75
• 6.2 INDEPENDENT-SAMPLES T-TEST 78
• 6.3 TESTING FOR MEDIATION 79
• 6.4 TESTING FOR MODERATION EFFECTS 81
• 7. DISCUSSION, IMPLICATIONS, LIMITATIONS, AND FUTURE RESEARCH DIRECTIONS 83
• 7.1 DISCUSSION OF THE RESULTS 83
• 7.2 THEORETICAL IMPLICATIONS 85
• 7.3 MANAGEMENT IMPLICATIONS 87
• 7.4 LIMITATIONS AND FUTURE RESEARCH DIRECTIONS 89
• 8. CONCLUSION 91
• REFERENCES 93