Does it matter who owns the AI device? Impact of device ownership on customer dissatisfaction in AI service failure

Artificial Intelligence (AI) is increasingly integrated into the service industry. However, AI service failures remain inevitable. Limited attention has been paid to how device ownership influences customers' emotional responses to AI service failures. This study is grounded in attribution theory and social impact theory, examining the impact of device ownership (customer-owned vs. merchant-owned) on customer dissatisfaction in AI service failures, along with its underlying mechanism and boundary conditions. Across four scenario-based experiments, the results demonstrate that customers experience lower dissatisfaction when service failures occur on their own devices rather than merchant-owned ones. This effect is mediated by responsibility attribution, such that using one's own device induces a greater internal attribution of failure. Nevertheless, this attributional process is significantly weakened when the AI is highly anthropomorphized or the service environment is crowded. These findings extend the literature on AI service failure by identifying device ownership as a crucial determinant of customer dissatisfaction, offering practical implications for mitigating negative responses.