生成式人工智能服务提供者版权侵权过错归责之明晰

Clarification of Liability for Copyright Infringement by Generative Artificial Intelligence Service Providers

在线阅读下载PDF

导出

摘要生成式人工智能技术的出现引发了版权侵权的风险,但我国法律目前并未就侵权责任的承担给出明确回应。可采用解释论的立场,结合现行法对该问题予以规制。生成式人工智能系统具有广泛应用范围及多重应用形式,不应认定其为产品;版权侵权与产品侵权侵害程度不同,故不可要求生成式人工智能服务提供者承担产品责任。服务提供者兼具提供技术支持与内容服务的双重属性,宜认定其为新型网络服务提供者。应适用关于网络服务提供者的归责原则,要求服务提供者承担一般过错责任,过错推定责任不具解释空间。在过错的认定上,可结合生成式人工智能生命周期构建体系:在建立阶段,进行数据处理并允许数据删除;在微调阶段,履行透明度义务;在应用阶段,对生成内容进行风险过滤与标识,可类推适用避风港规则。 The emergence of generative artificial intelligence(GAI)technology has precipitated risks of copyright infringement,yet current Chinese law remains silent on the allocation of liability for such infringements.An interpretative approach grounded in existing legal frameworks may be adopted to regulate this issue.Given GAI systems’broad applicability and operational complexity,they should not be classified as“products”.The distinction between copyright infringement and product liability in terms of harm severity further precludes extending product liability to GAI service providers.These providers,characterized by their dual role as both technical infrastructure suppliers and content facilitators,ought to be classified as a novel category of network service providers.Consequently,the doctrines of liability imputation applicable to network service providers shall govern,requiring GAI service providers to bear general fault-based liability,with no legal basis for presuming fault.To operationalize fault determination,a systemic framework aligned with the GAI lifecycle should be implemented:During the development phase,providers must process training data and permit deletion requests;in the fine-tuning phase,transparency obligations must be fulfilled;and at the application phase,risk filtering,content labeling,and analogous application of the“safe harbor”rules shall be mandated.

作者邓睿雯 Deng Ruiwen(College of Comparative Law,China University of Political Science and Law,Beijing 100088,China)

机构地区中国政法大学比较法学研究院

出处《中南财经政法大学研究生论丛》 2025年第1期53-63,共11页 Journal of the Postgraduates of Zhongnan University of Economics and Law

关键词生成式人工智能网络服务提供者版权侵权归责原则注意义务 Generative Artificial Intelligence Network Service Providers Copyright Infringement Doctrines of Liability Imputation Duty of Care

分类号 D923.41 [政治法律—民商法学]

作者简介邓睿雯(2001-),女,湖北武汉人,中国政法大学比较法学研究院2023级硕士研究生。