生成式人工智能的个人数据风险及其法律规制
Personal Data Risks of Generative Artificial Intelligence and Its Legal Regulation

以ChatGPT为代表的新一代生成式人工智能技术，具备自主学习性、强交互性、创新性等特点，对既有的个人数据风险治理体系构成了挑战。为应对其引发的个人信息安全风险，有必要从生成式人工智能的数据收集与处理、算法运行与其生成性内容三个关键要素进行探讨。针对个人数据侵权、算法偏见以及生成性内容的非法利用等问题，需要进行分类讨论。结合我国已有的人工智能立法实践分析，其对生成式人工智能的个人信息规制存在法律规范缺位、规则适用困难、治理手段乏力等现实难题，需从立法、法律适用、审查监督、司法救济四个层面提出相应的法律规制建议，以促进生成式人工智能的个人数据风险法律规制体系更加严谨和完善。
The new generation of generative AI technologies represented by ChatGPT, characterized by autonomous learning, strong interactivity, and innovativeness, poses a challenge to the established personal data risk governance system. In order to deal with the risk of personal information security caused by it, it is necessary to explore the three key elements of generative AI, namely data collection and processing, algorithm operation and its generative content. The issues of personal data infringement, algorithmic bias, and illegal utilization of generative content need to be discussed in a categorized manner. Combined with the analysis of China’s existing AI legislation and practice, the regulation of personal information of generative AI is characterized by the absence of legal norms, the difficulty of applying rules, and the lack of governance tools, etc. It is necessary to put forward the corresponding legal regulation proposals from the four levels of legislation, application of law, review and supervision, and judicial relief, so as to promote a more rigorous and perfect legal regulation system for the risk of personal data of generative AI.