生成式人工智能创作中的著作权风险及其化解  

Copyright Risks in Generative AI Creations and Their Mitigation

作　　者：贺琼琼 岳春晓 HE Qiongqiong;YUE Chunxiao(School of Law,Hubei university,Wuhan,Hubei,430062,China)

机构地区：[1]湖北大学法学院,湖北武汉430062

出　　处：《浙江树人学院学报》2025年第1期70-81,共12页Journal of Zhejiang Shuren University

摘　　要：生成式人工智能创作的技术过程可以分为数据收集阶段、训练阶段与生成阶段,其中蕴含复制权、改编权与信息网络传播权侵权风险。生成式人工智能创作在适用现有著作权法合理使用制度时面临困境。为平衡人工智能研发者、原著作权人以及社会公众的多方利益,有必要将生成式人工智能创作纳入法定许可。此种法定许可的适用主体应涵盖所有人工智能研发者,适用客体为已经发表的作品,适用条件不限定为非商业目的,适用行为包括复制与改编行为,且不允许著作权人事先声明排除适用。此外,还须基于作品市场价值制定明确的付酬标准,充分发挥著作权集体管理组织报酬转收的职能,并规定人工智能研发者的训练数据披露义务,通过一系列配套措施来保障生成式人工智能创作法定许可的顺利实施。The technological process of generative AI creations can be divided into three stages:data collection,training,and generation,each of which involves potential risks of copyright infringement,including reproduction rights,adaptation rights,and the right of communication through information networks.The application of existing copyright law's fair use doctrine to generative AI creations presents significant challenges.To balance the interests of AI developers,original copyright holders,and the public,it is necessary to bring generative AI creations under statutory licensing.This statutory license should apply to all AI developers as licensees,cover published works as the subject matter,and not be restricted to non-commercial purposes.The permitted acts should include reproduction and adaptation,and copyright holders should not have the right to preemptively exclude their works from the license.Additionally,a clear remuneration standard should be established based on the market value of works,levera-ging collective copyright management organizations to handle compensation collection and distribution.Furthermore,AI develop-ers should be required to disclose training data,ensuring the effective implementation of statutory licensing for generative AI crea-tions through a set of supporting measures.

关 键 词：生成式人工智能 著作权侵权 合理使用 法定许可 

分 类 号：D923.41[政治法律—民商法学] D922.17[政治法律—法学] D922.294