GAI服务提供者著作权间接侵权责任认定

Determining Liability for Indirect Copyright Infringement by GAI Service Providers

摘要: 生成式人工智能不具备法律主体资格，无法承担著作权侵权责任，其服务提供者不得以“技术中立”主张免责。生成式人工智能服务提供者与传统网络服务提供者差异显著，将界定其为新型网络服务提供者更为适宜。在归责原则的选择上，过错推定责任原则兼具合理性与可行性。服务提供者是否存在过错，关键在于其是否履行应尽义务，具体包括建立投诉举报渠道、风险提示、显著标识、技术性预防与训练数据合规审查义务。一旦服务提供者没有合理履行上述义务，即可认定其具有过错。明确侵权责任认定标准，不仅可为司法实践提供清晰的裁判标准，也为服务提供者设定可预期的合规边界，从而实现技术创新与著作权保护之间的动态平衡。

Abstract: Generative artificial intelligence lacks legal personality and cannot be held liable for copyright infringement; its service providers may not invoke “technological neutrality” to claim exemption from liability. As generative AI service providers differ significantly from traditional internet service providers, it is more appropriate to classify them as a new type of internet service provider. With regard to the choice of liability principle, the principle of presumption of fault is both reasonable and feasible. Whether a service provider is at fault hinges on whether it has fulfilled its obligations, which specifically include establishing channels for complaints and reports, providing risk warnings, displaying prominent notices, implementing technical safeguards, and conducting compliance reviews of training data. Should a service provider fail to reasonably fulfil the aforementioned obligations, it may be deemed to be at fault. Clarifying the criteria for determining liability for infringement not only provides clear adjudication standards for judicial practice but also establishes predictable compliance boundaries for service providers, thereby achieving a dynamic balance between technological innovation and copyright protection.