摘要
对人工智能版权侵权责任的分析应当摒弃"统合分析"模式,坚持"层次分析"方法,就深度学习中的复制行为与输出侵权作品的行为分别进行评价。就前者而言,基于复制权的式微及深度学习中的复制累积所形成的传播效应,应以是否合法获取复制件作为侵权判定的基础;就后者而言,现阶段不宜更改"接触+实质性相似"判定标准和过错推定原则,但应基于人工智能行为的自主性和算法黑箱,相应配置人工智能设计者、控制者"防抄袭"注意义务,并以公平责任作为补充。在技术推动变革的时代,应始终以建构主义的态度审视人工智能的发展,实现版权法与技术的良性互动。
The analysis of copyright infringement by artificial intelligence should abandon the integrated analysis mode and stick to hierarchical analysis,and evaluate the copying in advanced learning and the output of infringing works respectively.As for the former,because of the lowered status of reproduction right and the accumulation effect of reproduction in advanced learning,infringement analysis should be based on whether there exists lawful acquisition of the copyrighted material.As for the latter,it is still too early to adjust the"access+substantial similarity"standard and principle of presumption of fault.Based on the autonomy of AI and algorithm black box,the anti-piracy duty of care of AI designer and controller should be configured accordingly,and supplemented by fair liability.In a technology-driven world,people should always examine the development of AI with a constructivist attitude to realize the virtuous interaction between copyright law and technology.
出处
《北方法学》
CSSCI
北大核心
2021年第1期138-150,共13页
Northern Legal Science
基金
教育部哲学社会科学研究重大课题攻关项目“中美网络空间治理比较研究”(18JZD040)的阶段性成果。
关键词
人工智能
版权法
侵权责任
过错推定
artificial intelligence
Copyright Law
tort liability
principle of presumption of fault
