生成式人工智能背景下著作权侵权责任认定与治理路径研究

Research on the Identification of Copyright Infringement Liability and Governance Paths in the Context of Generative Artificial Intelligence

摘要: 生成式人工智能技术的发展正在改变知识生产与内容创造模式。以ChatGPT等大型语言模型为代表的生成式人工智能在文本、图像、音频等内容生成领域得到广泛应用，但其数据驱动、算法生成和输出不确定等特点也对传统著作权制度提出了挑战，涉及训练数据使用、生成内容权利属性、侵权责任认定以及服务提供者责任边界等问题。本文基于著作权法理论与人工智能治理视角，围绕人工智能生成内容作品认定、训练数据合理使用以及服务提供者侵权责任三个方面，分析生成式人工智能著作权治理中的现实困境，并提出完善生成内容评价标准、优化训练数据利用规则、构建以合理注意义务为核心的平台责任机制等治理路径，以期实现人工智能技术创新与著作权保护之间的平衡。

Abstract: The development of generative Artificial Intelligence (AI) is transforming the modes of knowledge production and content creation. Generative AI systems, represented by large language models such as ChatGPT, have been widely applied in the generation of text, images, audio, and other forms of content. However, their characteristics, including data-driven processing, algorithmic generation, and output uncertainty, have also posed challenges to the traditional copyright system, raising issues concerning the use of training data, the legal status of AI-generated content, the determination of copyright infringement liability, and the boundaries of liability for service providers. Drawing on copyright law theory and the perspective of AI governance, this paper examines the practical challenges of generative AI copyright governance from three dimensions: the copyrightability of AI-generated content, the fair use of training data, and the infringement liability of service providers. It further proposes governance approaches, including refining the criteria for evaluating AI-generated content, optimizing the rules governing the use of training data, and establishing a platform liability framework centered on reasonable duties of care, with the aim of striking a balance between technological innovation and copyright protection.