生成式人工智能平台的著作权侵权责任  

Copyright Infringement Liability of Generative Artificial Intelligence Platforms

作　　者：熊琦[1] Xiong Qi

机构地区：[1]华中科技大学法学院

出　　处：《环球法律评论》2025年第2期23-37,共15页Global Law Review

摘　　要：在生成式人工智能平台如何承担著作权侵权责任的问题上,由于我国相关治理文件中未能对提供内容与提供服务加以明确区分,致使平台在主体属性和归责事由上出现法源选择和适用的双重难题。根据域外和我国长期适用“避风港”规则的司法经验,法院一般以类推适用信息存储空间服务提供者版权侵权认定规则涵摄新兴网络服务提供者类型。虽然生成式人工智能平台在内容来源、生成过程和使用方式等方面都不同于传统的信息存储空间和搜索链接服务,但在著作权侵权责任的认定上,仍有必要类推适用。就主体定位而言,在现有法律体系内,生成式人工智能平台仍属于网络服务提供者的范围。对于事前必要措施和事后必要措施的解释,一方面当特定提示语和侵权内容之间出现稳定复现时,可将修改相关模型参数纳入必要措施范围;另一方面在延续适用“通知—必要措施”的基础上针对合规通知的侵权内容中涉及他人作品的部分,增加内容标注义务。The overlap between “AI Generated Content”(AIGC) and “User Generated Content”(UGC) enables online users to conveniently use artificial intelligence technology for creation, which sparks a debate on whether generative AI platforms should bear copyright infringement responsibility for the content they generate. In addressing how generative AI platforms should assume liability for copyright infringement, relevant regulatory documents in China do not clearly distinguish between providing content and providing services. This lack of clarity has resulted in dual challenges for platforms in selecting and applying legal sources regarding their subject status and grounds for liability. Drawing on judicial experience from both foreign jurisdictions and China's long-standing application of the “safe harbor” rule, courts typically analogize copyright infringement determination rules for information storage space service providers to encompass emerging types of online service providers. Although generative AI platforms differ from traditional information storage and search link services in terms of content source, generation process, and usage mode, it is still necessary to refer to network service infringement precedents with similar legal effects when determining their copyright infringement liability. Generative AI platforms should be included in the scope of network service providers to avoid concept confusion and systemic conflicts arising from the direct translation of technical concepts into legal concepts, and to address new issues by interpreting necessary ex ante and ex post measures. Based on established interpretative rules and judicial precedents for information storage space services and search link services, the elements for determining infringement by generative AI platforms can be addressed through the interpretation of “constructive knowledge-necessary measures” and “notice-necessary measures”. Regarding “constructive knowledge-necessary measures”, whether a generative AI platform con

关 键 词：生成式人工智能 网络服务提供者 “避风港”规则 通知—必要措施 

分 类 号：D923.41[政治法律—民商法学]