生成式人工智能绘画著作权侵权与规制路径研究  

A Study on Copyright Infringement and Regulation Paths for Generative Artificial Intelligence Painting

作　　者：王卓群 Wang Zhuoqun(School of Artificial Intelligence and Law,Southwest University of Political Science and Law,Chongqing 401120)

机构地区：[1]西南政法大学人工智能法学院,重庆401120

出　　处：《西部学刊》2025年第7期75-82,共8页Journal of Western

基　　金：重庆市社会科学规划项目“数字经济时代数据权属配置法律制度研究”(编号:2020YBFX44)的阶段性成果。

摘　　要：生成式人工智能绘画快速发展并普及,随之而来的是著作权侵权风险,其社会影响广泛。生成式人工智能绘画的技术特征包括具备深度学习能力与依托海量训练数据。结合国内外既有司法案例,可将其著作权侵权分为两类,即训练数据对在先著作权人的侵权,以及生成式人工智能绘画面临的盗用等侵权行为。当前在规制其著作权侵权的过程中存在诸多问题,包括生成式人工智能绘画著作权认定的不明确、训练数据集侵权案件的举证困难、训练数据集缺乏有效治理等。为有效规制其侵权行为,平衡技术创新和法律规制,应以使用者独创性贡献认定生成式人工智能绘画著作权,探索训练数据侵权案件中的举证责任倒置规则,加强训练数据来源安全与内容的安全治理。Generative artificial intelligence(AI)painting has developed rapidly and become widespread,bringing with it the risk of copyright infringement,which has a broad social impact.The technical characteristics of generative AI painting include deep learning capabilities and reliance on massive amounts of training data.Combining domestic and international judicial cases,copyright infringement can be divided into two categories:infringement of existing copyright holders by training data,and infringement faced by generative AI works,such as misappropriation.There are many difficulties in the process of regulating copyright infringement,including ambiguity in determining copyright ownership of generative AI works,evidentiary difficulties in training data infringement cases,and the lack of effective governance of training datasets.To effectively regulate infringement and balance technological innovation with legal regulation,it is proposed to identify AI painting copyrights based on the original contributions of users,explore the rule of reversed burden of proof in training data infringement cases,and strengthen the governance of the source and content of training data.

关 键 词：生成式人工智能绘画 著作权 著作权侵权 训练数据 

分 类 号：D923.41[政治法律—民商法学]