摘要
深度伪造是利用深度学习算法,实现视频、音频、图像的伪造和模拟形成虚假内容的总称。其危害性在于:深度学习技术背景下个体的广泛参与,深度伪造防不胜防;过度攫取公民隐私,信息泄露引发的系统安全风险高;社交网络中的暗黑心理驱使深度伪造内容的广泛传播;显性与隐性危害后果加持,辟谣难以修复。其法律规制困境是:技术风险的不确定性与法律规制的确定性之间的紧张关系愈发激烈;技术伦理的刚性制度约束欠缺,多元价值冲突难以调和;法律否定性评价的量化处罚形式无法达成罪责刑相适应。深度伪造平台化治理的正当性在于:规范获取渠道影响力与内容审核责任一致性;代理式监管的便利考量;主导人工智能治理话语体系的必然要求。应依托于内容发布平台的治理,形成事前的严格审核,事中的监控巡查、减少消除危害,事后的法律惩戒,联合外部约束与内部自律机制,构建深度伪造治理的新格局。
Deepfake is the use of deep learning algorithms to realize the forgery and simulation of video,audio,and images to form false content.Its harmfulness lies in:the extensive participation of individuals in the context of deep learning technology,and deepfake cannot be prevented;excessively grabbing citizens'privacy,the system security risk caused by information leakage is high;the dark psychology in social networks drives the widespread dissemination of deeply faked content;the explicit and implicit harmful consequences are exacerbated,and it is difficult to repair the rumor.The dilemma of its legal regulation is:the tension between the uncertainty of technical risk and the certainty of legal regulation;the rigid institutional constraints of technical ethics are lacking,and the conflict of multiple values is difficult to reconcile.The quantitative punishment form of legal negative evaluation cannot achieve the adaptation of crime,responsibility and punishment.The legitimacy of deep counterfeiting platform governance lies in:standardizing the influence of access channels and the consistency of content audit responsibilities;the convenience consideration of agency supervision;the inevitable requirement of leading the discourse system of artificial intelligence governance.We should rely on the governance of the content publishing platform to form a new pattern of deepfake governance by strict review in advance,monitoring and inspection in the process,reducing and eliminating hazards,legal punishment afterwards,and combining external constraints and internal self-discipline mechanisms.
作者
孙天宇
Sun Tianyu(Guangdong Police College,Guangzhou Guangdong 510440)
出处
《警学研究》
2023年第6期25-35,共11页
Police Science Research
基金
广东省普通高校青年创新人才类项目“公安大数据治理体系建构研究”的研究成果,项目编号:2019WQNCX092。
关键词
深度伪造
过程解构
平台化治理
法律规范
模式建构
deepfake
process deconstruction
platformized 1 governance
legal norms
model construction
