生成式人工智能著作权保护的路径转型

Path Transformation of Copyright Protection for Generative Artificial Intelligence

摘要: 生成式人工智能(AIGC)的爆发式增长正在挑战著作权法以“人类作者”为核心的传统预设，使得“全有或全无”的保护模式愈发难以适用。为此，本文提出“生成过程分层认定法”，将独创性判断从“结果导向”转向“过程导向”，从数据输入阶段的“人类意图注入”、算法运行阶段的“参数干预空间”到输出阶段的“表达形式可识别性”三个层面加以考察。在此基础上，构建“贡献要素权重参考框架”，将数据、算法、提示词和后期编辑纳入权利分配体系。此外，本文还探讨了过程导向方法在司法实践中的证据问题，设计了日志记录、第三方存证、专家辅助人及举证责任分配等配套制度。在技术治理方面，本文正视区块链技术的局限性，提出分层披露、智能合约审查等法律对策。整体上，本文主张著作权保护从“权利中心”转向“行为规制”，以推动形成更加系统的治理方案。

Abstract: The explosive growth of generative artificial intelligence (AIGC) is challenging the traditional “human author” premise of copyright law, making the “all-or-nothing” protection model increasingly difficult to apply. To address this issue, this paper proposes a “layered approach to generative process determination”, shifting the originality assessment from a “result-oriented” to a “process-oriented” framework. The analysis proceeds along three dimensions: “human intent injection” during the data input stage, “parameter intervention space” during algorithmic operation, and “identifiability of expressive form” during the output stage. On this basis, a “contribution factor weight reference framework” is constructed to incorporate data, algorithms, prompts, and post-editing into the rights allocation system. Furthermore, this paper examines the evidentiary issues of the process-oriented approach in judicial practice, designing supporting mechanisms such as logging standards, third-party preservation, expert assistants, and burden of proof rules. In terms of technological governance, this paper confronts the limitations of blockchain technology and proposes legal countermeasures including layered disclosure and smart contract review. Overall, this paper advocates a shift in copyright protection from a “rights-centered” to a “conduct-regulation” approach, promoting the formation of a more systematic governance scheme.