数据抓取行为的反不正当竞争法规制——以“三重授权原则”的适用为视角

The Regulation of Data Scraping under Anti-Unfair Competition Law—From the Perspective of the Application of the “Triple Authorization Principle”

摘要: 生成式人工智能的快速发展对传统数据抓取规制提出挑战，起源于“微博诉脉脉案”的“三重授权原则”在司法实践中不断扩张适用，从OpenAPI接口延伸至爬虫技术、从个人信息扩展至非个人信息、从获取环节覆盖至使用环节，在保护平台投入的同时，与数据流通政策、个人信息可携权及竞争自由产生张力。人工智能训练所需的海量公开数据难以实现事先授权，使该原则面临适用困境。应回归反不正当竞争法的行为规制本质，构建类型化规则：根据数据是否可识别及是否衍生，区分用户同意与平台授权的不同配置；增设人工智能训练例外条款，明确公开数据合理使用边界；引入透明度义务与合理使用抗辩，实现数据流通、权益保护与技术创新的三元平衡。

Abstract: The rapid development of generative AI challenges the traditional regulation of data scraping. The “Triple Authorization Principle”, originating from the Sina Weibo v. Maimai case, has been expansively applied in judicial practice—from OpenAPI to crawling technologies, from personal to non-personal information, and from data acquisition to utilization. While protecting platform investments, this expansion conflicts with data circulation policies, the right to data portability, and competitive freedom. AI training requires massive public data that can hardly obtain prior authorization, rendering the principle difficult to apply. The solution lies in returning to the behavioral regulation nature of anti-unfair competition law and establishing a typological framework based on data attributes. For identifiable primary data, only user consent is required; for non-identifiable public primary data, no platform authorization is needed; for identifiable derivative data, dual authorization is required; for non-identifiable derivative data, only platform authorization is required. Additionally, an exception clause for AI training, transparency obligations for data sources, and a fair use defense should be introduced to balance data circulation, rights protection, and technological innovation.