Abstract: This study develops a theoretical framework to explain the dynamic interplay among employees' professional identity threats, impression management behaviors, and the legitimacy construction of artificial intelligence (AI) tools within organizational contexts. As AI tools such as DeepSeek and ChatGPT become widely integrated into organizational practices due to their remarkable capabilities in innovation generation, decision-making support, and information analysis, understanding employees' responses and adaptations to these technological advancements becomes essential. Despite the notable improvements in efficiency and output quality facilitated by AI tools, employees frequently engage in impression management behaviors aimed at strategically modifying or obscuring the contributions of these technologies. Such behaviors reflect deeper anxieties regarding threats to their professional identity, particularly concerning perceived risks of diminished professional status, competence, and decision-making autonomy. Addressing existing theoretical gaps, this research integrates identity theory, impression management theory, and organizational legitimacy theory into a comprehensive, interactive theoretical framework. Prior studies often examined these theories independently, neglecting the nuanced interplay at the micro-level between individual cognition, behavior, and broader organizational legitimacy processes. This study systematically addresses this gap by posing four critical research questions: (1) Under what specific organizational and task-related circumstances do employees perceive threats to their professional identity upon the introduction of AI tools? (2) How do distinct perceptions of identity threats—status threats, competence threats, and autonomy threats—influence employees' specific impression management behaviors? (3) How do these impression management behaviors, in turn, influence the cognitive, pragmatic, and normative legitimacy construction of AI tools within organizations? (4) How does the legitimacy constructed around AI tools reciprocally influence employees' ongoing identity threat perceptions and subsequent behaviors, thereby creating a continuous feedback loop? To address these questions, the study first identifies and elaborates on key contextual factors that evoke perceptions of identity threats among employees. These factors are categorized into three main dimensions: technological attributes, task characteristics, and organizational contexts. Technological attributes include AI tools' decision transparency and autonomous decision-making capabilities, directly influencing employees' perceptions of control and their professional roles. Task characteristics, such as task specialization, uniqueness, and clarity of performance evaluation criteria, also significantly impact employees' feelings of professional competence and security. Meanwhile, organizational contexts—particularly managerial communication strategies emphasizing AI's efficiency and potential substitutability, alongside the organization's information feedback mechanisms—play critical roles in shaping employees' identity threat perceptions. Subsequently, the study explicates how these perceived identity threats motivate employees toward specific impression management strategies, including downplaying the contributions of AI tools, extensively modifying AI-generated outcomes, and concealing detailed information about tool usage. These impression management behaviors are strategic responses aimed at safeguarding professional status, competence perception, and autonomy within organizational social evaluations. Importantly, the research highlights that these behaviors do not merely stem from identity threats but also reinforce the very perceptions of threat that initiated them, resulting in a reinforcing negative feedback loop. Further, the analysis elucidates how employees' impression management behaviors directly affect the construction of AI legitimacy within organizations. Through deliberate distortion and concealment of information, employees' behaviors adversely influence colleagues' accurate understanding and evaluation of AI tools' cognitive legitimacy (perceived necessity and value), pragmatic legitimacy (actual utility and effectiveness in practice), and normative legitimacy (alignment with organizational culture and ethical standards). Consequently, these behaviors significantly obstruct the institutionalization and effective integration of AI technologies. The reciprocal dynamics between AI legitimacy construction and employees' identity threat perceptions are thoroughly explored, emphasizing that insufficient legitimacy enhances employees' perceptions of professional vulnerability, thus exacerbating further impression management behaviors and creating a persistent negative feedback cycle. In conclusion, this research contributes significantly to theoretical development by providing an integrative, interactive theoretical model that bridges the micro-level psychological and behavioral responses of employees with macro-level organizational legitimacy processes. By explicitly articulating how identity threats, impression management behaviors, and legitimacy construction dynamically interact, the study advances identity theory, impression management theory, and legitimacy theory, offering a more nuanced understanding of technology integration within organizational contexts. Practically, this study provides actionable strategies for organizational leaders to effectively manage the intricate psychological and behavioral challenges accompanying AI adoption, facilitating smoother institutionalization and long-term acceptance of transformative technologies.

摘要： 研究基于员工认知—行为互动的视角，深入探讨了组织引入人工智能（Artificial Intelligence，AI）工具过程中员工感知的专业身份认同威胁、印象管理行为以及AI合法性建设之间的互动关系与机制。研究首先明确了技术属性、任务特征和组织环境作为引发员工身份威胁感知的重要情境因素，并提出了专业地位威胁、专业能力威胁和自主性威胁三种具体的身份威胁感知类型，详细阐明了不同威胁感知如何驱动员工实施特定的印象管理行为，以及该行为如何产生认知强化作用。同时，研究进一步揭示了员工实施的印象管理行为不仅影响AI工具在组织内部认知、实践和规范三个层面的合法性建设，而且AI工具合法性建设程度也会反过来影响员工的身份认同威胁感知与印象管理行为，从而形成互动反馈机制。研究在理论上整合并深化了身份认同理论、印象管理理论与组织合法性理论的理解，也为管理实践提供了明确的策略指导，帮助组织有效应对AI工具应用过程中复杂的员工认知、行为与合法性管理挑战。