技术黑箱与规则重构:刑事诉讼中深度伪造证据的审查困境及路径选择

Technical black box and rule reconstruction:review dilemmas and path selection of deep-fake evidence in criminal litigation

在线阅读下载PDF

导出

摘要随着生成式人工智能技术的快速发展,深度伪造技术在刑事诉讼中的滥用对传统证据审查体系构成严峻挑战。技术黑箱特性导致证据真实性存疑、算法与司法审查权的结构性冲突以及现行法律规范与技术发展的脱节。深度伪造证据通过高度仿真的视听内容突破传统伪证识别逻辑,引发真实性审查标准模糊、技术依赖症加剧司法权让渡、证明责任分配失衡及证据链断裂等系统性风险。为应对上述风险挑战,亟须构建“制度—技术”协同治理框架:推动深度伪造技术专项立法,确立高风险场景下的强制标识与溯源义务;设计分层式审查模式,强化生成环境日志备案与算法透明度要求;创新证明责任分配机制,建立“技术质疑触发举证转移”规则;完善人工智能服务商行业规范,实施全链条算法审计制度;引入专家辅助人与技术调查官协同审查机制,弥补司法技术认知鸿沟。 With the rapid development of generative artificial intelligence technology,the abuse of deep-fake technology in criminal litigation poses a severe challenge to the traditional evidence review system.The characteristics of the technical black box lead to doubts about the authenticity of evidence,the structural conflict between the algorithm and the judicial review power,and the disconnection between the current legal norms and the technological development.Deep-fake evidence breaks through the traditional false evidence recognition logic through highly realistic audio and visual content,triggering systematic risks such as blurred authenticity review standards,intensified judicial power transfer due to technology dependence,unbalanced distribution of the burden of proof,and breaking of the evidence chain.To deal with the above mentioned risk challenges,it is urgent to construct a“system-technology”collaborative governance framework:promote special legislation on deep-fake technology,establish the mandatory marking and traceability obligations in high risk scenarios;design a hierarchical review model,strengthen the requirements for log filing of the generation environment and algorithm transparency;innovate the burden of proof distribution mechanism,establish the rule of“technical doubt triggering the transfer of proof”;improve the industry norms of artificial intelligence service providers,implement the whole chain algorithm auditing system;introduce the collaborative review mechanism of expert assistants and technical investigators to make up for the judicial technical cognitive gap.

作者潘金贵杨洪彬 Pan Jingui;Yang Hongbin(Southwest University of Political Science and Law,Chongqing 401120,China)

机构地区西南政法大学

出处《辽宁公安司法管理干部学院学报》 2025年第4期69-78,共10页 Journal of Liaoning Administrators' College of Police and Justice

基金 2025年重庆市研究生科研创新项目“数字检察视域下涉案财产全过程同步审查机制研究”(项目编号:CYS25236)的阶段性研究成果。

关键词深度伪造证据审查技术黑箱算法审计 deep-fake evidence review technical black box algorithm audit

分类号 D924.3 [政治法律—刑法学]

作者简介潘金贵(1973-),男,贵州毕节人,西南政法大学法学院教授、博士生导师,刑法学博士后,主要从事刑事诉讼法学、证据法学研究;杨洪彬(2002-),男,陕西安康人,西南政法大学2024级刑事诉讼法学硕士研究生,主要从事刑事诉讼法学、证据法学研究。