[目的/意义]生成式人工智能的广泛应用催生了多重数据安全风险，传统的回应型治理和集中式治理模 式显然难以应对新挑战，而敏捷治理则以其灵活的姿态彰显了独特的优越性。[方法/过程]首先，梳理生成式人工 智能在数据安全领域引发的多重风险，进而对三类治理模式在风险管理上的成效进行对比分析，在此基础上提出 我国应当尽快实现治理模式转变。然后以此为指导，构建具体的治理制度。[结果/结论]敏捷治理模式凭借其适 应性、柔韧性和包容性特质，为应对生成式人工智能等新兴技术的数据安全风险提供了有效的方案，能够在实践 中不断优化生成式人工智能数据安全治理体系。在敏捷治理模式下，应当树立“预防与应对并重”的适应性治理 理念，构建“多元参与，合作互动”的韧性治理机制，运用“技术叠加法律”的包容性治理工具，从而形成综合的生成 式人工智能数据安全治理体系。

[Purpose/significance] The widespread use of generative artificial intelligence has given rise to multiple da⁃ ta security risks. Traditional reactive and centralized governance models seem inadequate to address these new chal⁃ lenges. In contrast, agile governance, with its flexible approach, demonstrates a unique superiority. [Method/process] Firstly, a meticulous examination of the multifaceted risks in data security posed by generative artificial intelligence is conducted. Subsequently, a comparative analysis of the efficacy of three governance models in risk management is car⁃ ried out. It is suggested that China should promptly transition its governance model. Guided by this, a specific gover⁃ nance framework is constructed. [Result/conclusion] The agile governance approach, characterized by its adaptability, flexibility, and inclusiveness, offers a highly effective strategy for addressing data security risks associated with emerg⁃ ing technologies like generative artificial intelligence. It allows for the continuous optimization of the governance sys⁃ tem for generative artificial intelligence data security. Within the agile governance framework, it is essential to embrace the adaptive governance concept of "equal emphasis on prevention and response", construct a resilient governance mechanism that promotes "diverse participation and collaborative interactions"，and employ inclusive governance ap⁃ proaches that combine technology with legal measures. This way, a comprehensive framework for ensuring the security of generative artificial intelligence data is constructed.