训练数据主体权益保护的新型数据财产权构建

Constructing A New Data Property Rights System to Protect the Rights and Interests of Training Data Subjects

生成式人工智能的三大要素包括训练数据、算力和算法,其中训练数据量决定着生成式人工智能的“想象力”和创作能力,但生成式人工智能处理训练数据进行模型训练之时也产生了对训练数据承载的主体权益的侵害,而且已成立的权利(权益)保护制度难以适应训练数据主体权益保护。为此,应当建立符合训练数据性质、结构和特征的新型数据财产权制度,以此实现对于生成式人工智能训练数据承载的各主体权益的保护,落实国家《数据二十条》“构建适应数据特征、符合数字经济发展规律、保障国家数据安全、彰显创新引领的数据基础制度”的指导思想,进而促进生成式人工智能的健康发展。

Training data is one of the three essential elements in generative artificial intelligence,alongside data,computing power,and algorithms.The quantity of training data significantly influences the“imagination”and creative capabilities of generative AI.However,the use of training data in model training by generative AI also raises concerns about the infringement of the rights and interests of training data subjects.The existing legal framework for protecting rights and interests struggles to adapt to the unique challenges posed by training data.Therefore,it is imperative to establish a novel data property rights system tailored to the nature,structure,and characteristics of training data.This will ensure the protection of the rights and interests of all stakeholders involved in the generation of generative AI.This aligns with the guiding principles of the national Twenty Data Regulations,which emphasize the need to construct a data foundational system that suits the characteristics of data,complies with the laws of digital economic development,safeguards national data security,and promotes innovation-led development.Ultimately,this initiative will contribute to the healthy development of generative artificial intelligence.