论人工智能生成数据法律保护的多元分层模式——兼评"菲林案"与"Dreamwriter案

根据人类是否介入或介入是否具有实质性影响,人工智能生成数据可分为辅助生成数据和自动生成数据.既有学术探讨中,对此的各种保护模式各有优劣,但多集中在一种保护模式的单项选择上.结合"菲林案"与"Dreamwriter案"的判决结果,可对生成数据采取多元分层的保护模式:对独创性的辅助生成数据,可纳入人类作品范畴获得著作权;对具客观"独创性"的自动生成数据,可设立新型邻接权——"数据生成者权"加以保护;辅助与自动生成中都存在不具"独创性"的数据,可采数据财产权保护模式.当数据生成者权、数据财产权尚未设立或设立后保护仍有遗漏的,反不正当竞争法可对人工智能生成数据予以兜底保护.

On the Multi-Layered Model of Legal Protection of Data Generated by Artificial Intelligence:A Review of"Film Case"and"Dreamwriter Case"

Data generated by AI (Artificial Intelligence) can be divided into two types:human-led data that uses AI as a tool to assist in generating data, and automated data generated with little human intervention and no credit to humans, or even no intervention completely. The "Filming Case" and the "Dreamwriter Case" have established two types of determination results-the auxiliary generated and original data are classified as human works; while for automatically generated data that is not "original", recognizing the value of their property and providing legal protection, but giving no positive response to other issues. In the academic discussion, although various protection modes have their advantages and disadvantages, they all focus on one single protection mode. In fact, a multi-layered protection model can be adopted for artificial intelligence to generate data. The original auxiliary generated data can be included in the category of human works to obtain copyright protection. And a new type of neighboring right-data generator right can be established to protect the automatically generated data with objective "ingenuity". If there is no original data in auxiliary generation and automatic generation, the value of their property can be protected by data property right. When neighboring rights and data property rights have not been established, or protection is still missing after establishment, the protection model of anti-unfair competition law can provide comprehensive protection for AI-generated data.