AI产品侵权归责原则的法经济学考察——兼评欧盟《人工智能责任指令》  

A Legal and Economics Review of AI Product Infringement Liability Principles--omments on the EU’s Artificial Intelligence Liability Directive

作　　者：蒋言斌[1] 高卉一 JIANG Yanbin;GAO Huiyi(School of Law,Central South University,Changsha,Hunan 410083,China)

机构地区：[1]中南大学法学院,湖南长沙410083

出　　处：《中州大学学报》2024年第3期39-46,共8页Journal of Zhongzhou University

基　　金：湖南省社会科学成果评审委员会重大课题“军民融合机制、模式及其选择研究”(XSP19ZDA002);中南大学校企联合项目“区域性学位论文数据库设计和作者利益分配系统”(2022XQLH021)。

摘　　要：AI产品是以AI技术为基础,以销售、使用为目的,依赖研发者设定的固定程序,通过计算进行决策并完成相应任务,被广泛应用于各行业的动产领域;AI产品侵权归责原则大致可分为共同责任侵权原则、严格责任原则和过错侵权原则三类。在适用侵权归责原则时,常见的困境包括共同侵权责任原则缺乏归责上的“共同性”,严格责任原则具有“片面性”,以及过错责任原则难以确定“过错”与“因果关系”。此外,责任主体尚不明确。基于社会成本与委托代理理论,AI产品侵权归责理论可从制度激励、信息不对称与风险分散三个维度解析归责的基本逻辑,以实现社会福利最大化。欧盟《人工智能责任指令》在一般条款外仅规定了举证责任、证据披露和因果关系推定等要素,对产品侵权问题的规制存在漏洞,它所确立的严格责任制度仍具有补充和提升空间。为实现欧盟《人工智能责任指令》下的社会福利最大化,尝试将过错责任原则适用于生产者,将剩余责任即严格责任原则适用于驾驶员,即若生产者存在过错,驾驶员和生产者将以各自的责任份额分担责任;如果生产者被证明遵循了应有的注意义务,那事故损失必须完全由被适用严格责任原则的驾驶员承担。AI products are based on AI technology for the purpose of sales and use.They rely on fixed programs set by developers to make decisions and complete corresponding tasks through calculations.They are widely used in the field of movable property in various industries;AI product infringement liability principles can be roughly divided into three categories:the principle of joint liability tort,the principle of strict liability and the principle of fault tort.When applying the principle of tort liability,common dilemmas include the lack of“commonality”in liability in the principle of joint tort liability,the“one-sidedness”of the principle of strict liability,and the difficulty in determining“fault”and“causal relationship”in the principle of fault liability.In addition,the subject of responsibility is not clear yet.Based on social cost and principal-agent theory,AI product infringement liability theory can analyze the basic logic of liability from the three dimensions of institutional incentives,information asymmetry and risk dispersion,in order to maximize social welfare.The European Union’s Artificial Intelligence Liability Directive(referred to as the“Directive”)only stipulates elements such as the burden of proof,evidence disclosure,and presumption of causality in addition to general provisions.There are loopholes in the regulation of product infringement issues,and the strict liability system established by the“Directive”still remains.There is room for supplementation and improvement.In order to maximize social welfare under the“Directive”,this article attempts to apply the strict liability,the principle of fault liability to producers and the principle of residual liability to drivers.That is,if the producer is at fault,the driver and the producer will be share responsibility in their respective shares of liability;if the producer is proven to have followed due care,the accident losses must be borne entirely by the driver to whom the strict liability rules apply.

关 键 词：AI产品侵权归责原则 法经济学 委托代理理论 社会成本理论 欧盟《人工智能责任指令》 

分 类 号：D923.7[政治法律—民商法学]