人工智能与著作权法的法律问题研究 — 结合中日经验的考察 — = Legal Issues of Artificial Intelligence and Copyright Law : A Comparative Study Based on Chinese and Japanese Experiences

다국어 초록 (Multilingual Abstract)

随着生成式人工智能技术的飞速发展，其在文本、图像等领域的大规模应用引发了诸多著作权法上的争议性问题。本文立足于“生成式AI生成内容是否构成著作物、应由谁享有权利、AI学习开发阶段使用他人著作物是否合法，以及使用AI产生生成物的侵权责任如何认定”等四大核心问题展开研究。 在方法上，本文以中国与日本的立法与实务为比较基础，考察两国在著作物属性认定、作者界定、权利限制制度、侵权判断标准等方面的制度差异与共同趋势。现行著作权制度普遍以“人类创作性”为前提，难以直接适用于AI独立生成内容，亟需通过解释论与制度设计对相关问题作出回应。进一步地，关于生成式AI在训练阶段中使用他人著作物的合法性边界，中日著作权权利限制制度存在不同规定，需结合日本文化厅解释性文件等内容进行分析。在使用生成式AI产生生成物的侵权责任层面，本文讨论了如何判断生成物的“依据性”与“类似性”，结合被称为AI生成使用阶段侵权世界第一案的“广州奥特曼”判决内容以及中国AI法《生成式人工智能服务管理暂行办法》进行考察。 在本文的前序部分，提及了日本于2025年5月正式颁布的《人工智能开发促进法》。该法规定了关于人工智能发展的基本方针，但并未设定实质性的规制措施。在本文的最后部分，指出人工智能不仅为著作权法带来了挑战，也带来了便利。其中具有代表性的，是中国法院正在应用的“版权AI智审”系统，该技术正在促进著作权侵权判断的效率与精准性。此外，除著作权领域外，人工智能在专利法中亦引发争议。本文简要介绍了包括日本法院近期否定AI可作为发明人申请专利的“DABUS案”判决，以及中国在AI能否作为发明人方面所展开的立法性讨论。通过中日经验的比较考察，本文旨在为建立既保障创作者权益、又促进技术发展的AI时代著作权规范提供参考。

다국어 초록 (Multilingual Abstract)

With the fast development of generative artificial intelligence (AI), its wide use in areas like text and image creation has raised several important copyright law issues. This article focuses on four key legal questions: (1) whether AI-generated content can be considered a copyrighted work; (2) who should own the rights to such content; (3) whether using others’ works in the training stage of AI is legal; and (4) how to determine whether the use of AI-generated content constitutes copyright infringement.
This article compares the laws and practices of China and Japan. It looks at how the two countries define “authorship” and “originality,” how they limit copyright, and how they judge copyright infringement. Most current copyright laws are based on the idea of human creativity. Because of this, they do not clearly apply to content made only by AI. As a result, legal interpretation and institutional adjustments are needed to respond to these challenges. For the training stage of AI, the article focuses on copyright limitations and exceptions. It compares China’s limitations with Japan’s approach, using guideline from Japan’s Agency for Cultural Affairs. For the use of AI-generated outputs, the article discusses how to judge “access” and “substantial similarity”. It also discusses the Ultraman case in China, which is considered the world’s first court decision on copyright infringement during the generative use phase of AI. In addition, it examines China’s AI legislation, the Interim Measures for the Management of Generative Artificial Intelligence Services, which sets out regulatory rules for generative AI services at the national level.
In the preliminary part of this article, reference is made to Japan’s official promulgation of the Act on the Promotion of Research，Development and Application of Artificial Intelligence-Related Technologies in May 2025. The Act sets out basic policy directions for AI development but does not contain substantive regulatory provisions. In the concluding part of the article, it is noted that AI brings not only challenges but also benefits to copyright law. A representative example is the “Copyright AI Smart Judgment” system currently being used by Chinese courts, which has contributed to improving the efficiency and accuracy of copyright infringement determinations. In addition, beyond the field of copyright, AI has also raised issues in patent law. The article briefly introduces the recent “DABUS” judgment in which a Japanese court denied the possibility of AI being recognized as an inventor, as well as legislative discussions in China concerning whether AI can be regarded as an inventor. By comparing China and Japan, this article aims to provide perspectives for building a copyright framework in the age of AI that both protects creators’ rights and promotes technological development.