生成式人工智能素材利用之侵权风险与应对
Infringement Risks and Responses in the Utilization of Generative AI Materials
摘要: 生成式人工智能(Generative Artificial Intelligence,以下简称GenAI)技术的广泛滥用,司法实践中所涉嫌的案例越来越多,鼓励技术创新与保护权利人利益之间产生冲突。GenAI基于对海量作品的学习和模仿,生成用户指令内容。然而,这一智能过程深陷著作权侵权风险与规制困境。首先,分析在素材获取阶段和算法训练阶段,可能涉及复制权、汇编权、翻译权等著作权权利的侵犯,详细阐述GenAI素材训练的关键环节,如数据抓取、数据清洗与标注等,将其逐一与复制权、汇编权、改编权进行对应分析;其次,分析现行著作权法在应对此类问题时的困境,即我国“合理使用”以及“法定许可”制度难以适配回应GenAI素材利用之侵权风险;最后,提出确立“GenAI素材训练合理使用适用之例外”,同时提出以“技术赋能”建立数据披露机制。本文旨在分析GenAI在素材获取和算法训练阶段中所面临的著作权侵权风险,审视现行著作权法体系规制框架的局限性,探寻在人工智能时代背景下,兼顾技术创新与权利保护的优化路径。
Abstract: The widespread abuse of Generative Artificial Intelligence (GenAI) technology has led to an increasing number of suspected cases in judicial practice, creating a conflict between encouraging technological innovation and protecting the interests of rights holders. GenAI generates user-instructed content based on learning and imitating a vast amount of works. However, this intelligent process is deeply entangled in the risk of copyright infringement and regulatory dilemmas. Firstly, this paper analyzes the potential infringement of copyright rights such as the right of reproduction, the right of compilation, and the right of translation during the stages of material acquisition and algorithm training. Secondly, it examines the predicament of the current copyright law in addressing such issues, namely, the difficulty of the “fair use” system in China to adapt to and respond to the infringement risks of GenAI’s material utilization. Finally, it proposes the establishment of an “exception for the application of fair use in GenAI material training” and suggests the establishment of a data disclosure mechanism through “technology empowerment”. This paper aims to analyze the copyright infringement risks faced by GenAI during the stages of material acquisition and algorithm training, review the limitations of the current copyright law regulatory framework, and explore an optimized path that balances technological innovation and rights protection in the context of the AI era.
参考文献
|
[1]
|
徐恩平. 生成式人工智能司法应用的风险与出路[J]. 中阿科技论坛(中英文), 2024(1): 149-155.
|
|
[2]
|
刘友华, 魏远山. 机器学习的著作权侵权问题及其解决[J]. 华东政法大学学报, 2019, 22(2): 68-79.
|
|
[3]
|
吴汉东. 知识产权总论[M]. 北京: 中国人民大学出版社, 2020.
|
|
[4]
|
Li, Z. (2025) Copyright Infringement Risks and Responses of Generative Artificial Intelligence. Science of Law Journal, 4, 23-32.
|
|
[5]
|
吴汉东. 人工智能生成作品的著作权法之问[J]. 中外法学, 2020, 32(3): 653-673.
|
|
[6]
|
刘禹. 机器利用数据行为构成著作权合理使用的经济分析[J]. 知识产权, 2024(3): 107-126.
|
|
[7]
|
赵旭. 生成式人工智能在机器学习中的合理使用问题[J]. 暨南学报(哲学社会科学版), 2024, 46(3): 79-95.
|
|
[8]
|
黄保勇, 施一正. 区块链技术在版权登记中的创新应用[J]. 重庆大学学报(社会科学版), 2020, 26(6): 117-126.
|