【作者】 王丽群；

【导师】 党玺；

【作者基本信息】 浙江理工大学 ， 法律（非法学）（专业学位）， 2023， 硕士

【摘要】 人工智能生成物在形式和逻辑上足以“以假乱真”,倘若人工智能被用于抄袭、洗稿等侵权行为,将危害个人利益和社会公共利益。一旦发生这样的情况,谁需要为人工智能创作的版权侵权问题承担责任?是为人工智能创作算法编程的设计者,还是为人工智能开发注入资金的生产者,抑或是实际使用人工智能创作的使用者?这也是文章所讨论的重点问题。为了明确各个主体如何分担人工智能的版权侵权责任,文章将根据人工智能创作的模式和特点展开进一步的研究。人工智能在数据采集、机器学习阶段都有可能产生版权侵权的风险,但是前两个阶段的风险行为最终能否被认定为侵权,仍然要结合输出阶段的人工智能生成物来判断。根据“接触+实质性相似”原则,人工智能生成物被认定为与他人的版权作品构成实质性相似,则人工智能的创作行为属于版权侵权。人工智能版权侵权责任规制中出现了诸多难题。一是无法使用传统的整体观感法、抽象分离法以及内外部测试法认定实质性相似。二是各类侵权情形下责任主体尚不明确。三是我国尚未出台人工智能的国家标准、行业标准,加之主体义务规定缺位,这导致被侵权人对因果关系的举证难度上升。最后,针对侵权规制中出现的难题分别提出解决方案。一是通过借助人工智能系统优化整体观感法、抽象分离法并引入市场替代法兜底的方式解决了实质性相似认定难的问题。二是引入区别对待归责的方法定位责任主体。因人工智能质量问题引发的侵权适用产品责任,由生产者或销售者承担无过错责任;因人工智能设计不当引发的侵权,由人工智能设计者承担过错责任;因人工智能使用不当引发的侵权,适用法人作品制度或承揽合同制度,由人工智能使用者承担过错责任;人工智能自主原因引发的侵权,适用公平责任,由使用者和被侵权人共同分担责任。三是通过明确各个利益主体的义务和减轻被侵权人举证负担的方式加强因果关系的认定。更多还原

【Abstract】 Computer-generated works are are similar to human works in form and logic.If artificial intelligence is used for plagiarism,manuscript laundering,etc.,it will damage personal interests and public interests.In a case like this,who needs to be responsible for the copyright infringement issues created by artificial intelligence? Is it a designer of AI programming,a producer investing in AI,or a user of AI?This is the key issue discussed in this paper.In order to clarify how various subjects share the copyright infringement liability of artificial intelligence,this paper will carry out further research according to the mode and characteristics of artificial intelligence creation.The risk of copyright infringement may arise in the data acquisition and machine learning stages of artificial intelligence.However,whether the risky behaviors in the first two stages can be ultimately identified as infringements still needs to be judged by combining the computer-generated works in the output stage.According to the rule of “access and substantial similarity”,if computer-generated works are deemed to be substantially similar in composition to the work of others,the creation behavior of artificial intelligence will constitutes copyright infringement.Many difficulties have emerged in the regulation of AI copyright infringement liability.First,traditional whole perception method,abstract separation method and extrinsic-intrinsic test method cannot be used for identification of substantial similarity.Second,the subject of liability in various types of infringement situations is not clear.Third,our country has not issued national standard,industry standard,and the stipulation of subject obligation is absent,which leads to increased difficulty for victims to prove causality.Finally,the paper proposes solutions to the problems in tort regulation.First,by optimizing the whole perception method,abstract separation method and introducing market substitution method with the help of artificial intelligence system,the problem of substantive similarity identification is solved.The second is to introduce the different treatment of imputation to locate the subject of responsibility.Producers or sellers shall bear no-fault liability for the products liability caused by the quality problems of artificial intelligence;In case of infringement caused by improper design of artificial intelligence,the artificial intelligence designer shall bear the fault liability;In case of infringement caused by improper use of artificial intelligence,the system of works made for hire or the system of undertaking contract shall be applied,and the user of artificial intelligence shall bear the fault liability;The infringement caused by artificial intelligence autonomy shall be subject to fair liability,which shall be shared by the user and the victim.The third is to strengthen the identification of causality by clarifying the obligations of each interest subject and reducing the burden of proof of the victim.更多还原