生成式人工智能的性质及其生成物权利归属

The Nature of Generative Artificial Intelligence and the Attribution of Rights to Its Outputs

摘要: 我国现行法律尚未对生成式人工智能(GAI)及其生成物的法律属性作出明确界定。随着ChatGPT的流行，社会关注度持续升温，厘清其性质已成为保障技术应用与推广的法律基础之需。GAI的运行机理本质上不同于人类思维模式，亟需就其本身及生成物的应用规范进行分析。GAI的显著特征在于其非人为干预的自主性，而著作权法的规制目标、价值体系及权利结构均以“人”为核心。其自主性使其难以被简单归类为传统意义上的创作工具。当前阶段，GAI既非适格的法律主体，也非著作权法框架下纯粹的“工具”。贸然突破著作权法基于“人”本位的立法范式边界，为GAI创设一个特定的“人”的法律地位，不仅有悖立法传统，且因缺乏对其权责配置的系统性研究，法律也难以要求GAI承担相应责任。鉴于此，将GAI生成物的权利归属直接归于特定法律主体存在障碍。相较之下，原则上将其纳入公有领域，并辅以“署名推定”规则规范特定使用情境，更能契合著作权法激励创新创作的价值目标。

Abstract: China’s current laws have not clearly defined the legal attributes of generative artificial intelligence (GAI) and its outputs. With the popularity of ChatGPT, social attention has been continuously rising, and clarifying its nature has become a necessary legal basis for ensuring the application and promotion of this technology. The operational mechanism of GAI is essentially different from human thinking patterns, so it is urgent to analyze the application norms for GAI itself and its outputs. A significant characteristic of GAI lies in its autonomy without human intervention, while the regulatory objectives, value system, and right structure of copyright law all take “humans” as the core. Its autonomy makes it difficult to be simply classified as a traditional creative tool. At the current stage, GAI is neither a qualified legal subject nor a pure “tool” under the framework of copyright law. Rashly breaking through the boundaries of the “human-oriented” legislative paradigm in copyright law and creating a specific “human” legal status for GAI is not only contrary to legislative traditions, but also makes it difficult for the law to require GAI to bear corresponding responsibilities due to the lack of systematic research on its rights and obligations configuration. In view of this, there are obstacles to directly attributing the rights of GAI outputs to specific legal subjects. In comparison, incorporating them into the public domain in principle, supplemented by the “presumption of authorship” rule to regulate specific usage scenarios, can better align with the value goal of copyright law in encouraging innovative creation.