Abstract: Responsible innovation (RI) provides a new solution to artificial intelligence (AI) governance, yet it is still an early stage for the research on employees’ RI behavior. This paper builds an integrated TPB-ANM theoretical model to analyze the influence factors of AI employees’ RI behavior from both self-interest and altruism perspectives, and adopts the structural equation model (SEM) to carry out an empirical test. The research findings are as follows: (1) Employees’ RI attitudes, injunctive norms, descriptive norms, personal norms and perceived behavior control have positive influence on employees’ RI willingness, and injunctive, descriptive and personal norms further influence employees’ RI behavior through the mediating role of employees’ RI willingness. (2) Injunctive, descriptive and personal norms have direct influences on employees’ RI behavior, among which injunctive norms has the most significant impact. (3) Consequential awareness has a positive bearing on ascription of responsibility and employees’ RI attitudes, while ascription of responsibility has a positive influence on personal norms. Based on the above research findings, this paper proposes constructive suggestions from the dimensions of the government, enterprises and employees to improve employees’ RI enthusiasm and behavioral initiative and release employees’ RI vitality. Hopefully, the paper can be a useful reference for the RI development of AI firms.

Key words: employees’ responsible innovation behavior, theory of planned behavior (TPB), norm activation model (NAM), AI