生成式人工智能服务提供者版权注意义务的构建

Construction of Copyright Duty of Care for Generative AI Service Providers

摘要: 生成式人工智能在内容生产中的深度参与重构了信息生产传播秩序，变相成为内容共创者与持续系统性侵权风险把关者。在主要靠被动的以个案为中心事后反应与平台行为谨慎的传统网络服务提供者的分类体系下，人工智能服务提供者容易陷入“过度防御”与“责任逃逸”。基于网络服务提供者侵权责任理论，结合生成式人工智能的技术控制力与持续侵权风险，依据作品类型的可识别性，应构建以风险分层和参与程度为核心的差异化注意义务规范体系，在版权权利保护、平台责任，用户生成式人工智能注意义务网络服务提供者平台责任侵权责任创作自由间实现平衡。

Abstract: The deep involvement of generative artificial intelligence in content production has the order of information production and dissemination, effectively transforming AI into a content co-creator and a gatekeeper of continuous systemic infringement risks. Under the traditional classification system of network service providers, which relies primarily on passive, case-centered ex post reactions and cautious platform behavior, AI service providers are prone to fall into either “over-defense” or “liability evasion.” Drawing on the tort liability theory of network service providers, and considering the technological and risk mechanisms of generative AI, a differentiated normative system of duty of care should be established based on risk stratification and the degree of participation, in light of the identifiability of work types. This system aims to achieve a balance among copyright protection, platform liability, and user creative freedom.