Stealing Fire or Burning Out? The Dual Impact of Generative AI on Employee Bootleg Innovation

The rapid adoption of generative AI, exemplified by ChatGPT, has sparked intense debate regarding its dual impact on organizational behavior. While it promises enhanced productivity, it also introduces significant cognitive and adaptational demands. Addressing this tension, this study investigates the impact of generative AI application on employee bootleg innovation, which is defined as a form of concealed and unauthorized innovation. Moving beyond the traditional focus on formal innovation, we employ the Job Demands-Resources (JD-R) model to uncover a dual-path mechanism. Based on multi-phase survey data from 305 employees, the findings reveal that generative AI acts as a unique job resource that facilitates concealed and independent ideation, thereby promoting bootleg innovation by enhancing work engagement. Conversely, acting as a job demand, the cognitive load and vigilant burden associated with AI application increase job burnout, which subsequently inhibits bootleg innovation. Furthermore, entrepreneurial leadership, functioning as a critical resource within organizations, moderates these pathways. It strengthens the positive effect of generative AI on work engagement and buffers the cognitive strain leading to job burnout. This moderation extends to the indirect effects, enhancing the gain path and weakening the loss path. Overall, this study highlights the “double-edged sword” nature of generative AI, providing valuable theoretical and practical insights for managing emergent and concealed innovation behaviors within AI-integrated work environments.