摘要
[研究目的]在生成式人工智能快速发展的背景下,深度伪造技术(Deepfake)已从数字娱乐工具演变为具有显著级联效应和系统性危害的新兴社会风险源。其影响贯穿微观个体权益、中观社会秩序乃至宏观国际政治格局,形成跨领域、跨层级的风险扩散路径。因此,通过系统分析国际经验与本土实践,对于完善我国深度伪造治理举措,构建兼顾安全与发展的治理体系具有重要意义。[研究方法]通过构建“技术—制度—社会”三维分析框架,系统阐释深度伪造级联风险的生成机制与演化路径,基于比较研究法与案例分析法,聚焦美国、欧盟和中国三大主要行为体,从技术研发、内容创作、信息流通与社会协同四个维度,梳理其在深度伪造全生命周期治理中的策略差异与实践特征。对比不同监管框架的优劣,识别共性困境与制度瓶颈,为优化治理路径提供实证基础。[研究结论]美国、欧盟和中国三大行为体治理路径虽各有侧重,但均面临级联风险识别滞后、规则碎片化、治理效能碎片化等问题。我国应通过创新“技术嵌入型”治理范式,推动检测识别、溯源阻断与合规科技协同发展;完善制度体系,细化平台规制,强化事后救济;构建政府主导、平台尽责、行业自律、公众参与的多元协同共治体系等方式,实现深度伪造技术的“善治”。
[Research purpose]Amid the development of Generative Artificial Intelligence(GAI),Deepfake has transformed from an entertainment tool into an emerging societal risk with cascading systemic hazards.Its impact spans from individual rights and social order to the international political landscape,forming a cross-domain,multi-level risk pathway.Therefore,systematically analyzing international experience and local practices is crucial for refining China’s Deepfake governance and building a system that balances security and development.[Research method]Using a“technology-institution-society”analytical framework,this study examines the generation mechanism and evolutionary path of deepfake cascade risks.Based on comparative research and case analysis methods,this study focuses on three major actors,the United States,the European Union,and China,and examines their strategic differences and practical characteristics in the governance of deepfakes across the entire lifecycle,from four dimensions:technology development,content creation,information dissemination,and social collaboration.By comparing the strengths and weaknesses of different regulatory frameworks,it identifies common challenges and institutional bottlenecks,providing an empirical foundation for optimizing governance pathways.[Research conclusion]Although the governance approaches of the three major actors—the United States,the European Union,and China—each have their own focus,they all face challenges such as delayed identification of cascading risks,fragmented rules,and fragmented governance effectiveness.China should promote the coordinated development of detection and identification,traceability and blocking,and compliance technology by innovating a“technology-embedded”governance paradigm,improving the institutional system,refining platform regulations,and strengthening post-event remedies,and building a multi-stakeholder collaborative governance system led by the government,with platforms fulfilling their responsibilities,industry self-discipline,and public participation,to achieve“good governance”of deepfake technology.
作者
令宇豪
张文俊
李若源
Ling Yuhao;Zhang Wenjun;Li Ruoyuan(Shaanxi Branch of the National Computer Network Emergency Response Technical Team/Coordination Center of China,Shaanxi,Xi’an,710075;Stockholm University,School of Law,The Kingdom of Sweden,Stockholm,SE-10691)
出处
《科技智囊》
2025年第11期63-69,共7页
Think Tank of Science & Technology
关键词
深度伪造
监管
级联风险
全生命周期
风险治理
人工智能
Deepfake
supervision
cascade risk
the whole life cycle
risk governance
artificial intelligence
