论虚拟数字人应用的数据安全法律保障——以生成式人工智能实施为进路

On the Legal Protection of Data Security for Virtual Digital Human Applications——With Generative Artificial Intelligence Implementation as a Way Forward

作为生成式人工智能的典型代表之一,虚拟数字人基于大数据模型训练进行文本、图片、影音算法生成,并应用于文化旅游、新闻娱乐、电商游戏等场景。新型数据安全风险也随之产生,如非法获取与利用数据风险、数据偏见与数据存储风险、虚假生成内容与不正当竞争风险、刑事风险等。由此,面对数据安全与数据利用原则之间的现实张力,坚持并实施数据安全的统筹安全与发展原则数据利用的分类分级原则,秉持“人身—财产”差异性数据分类模式,实现虚拟数字人数据“识别—溯源—监管”的合规路径,从而明确数据采集的范围、方式,厘清企业与政府之间对数据安全风险的监管界限。

As one of the typical representatives of generative artificial intelligence,virtual digital people are trained based on big data models for text,picture,audio and video algorithm generation,and are applied to cultural tourism,news and entertainment,e-commerce games and other scenarios.New types of data security risks have also arisen,such as the risk of illegal access and utilization of data,the risk of data bias and data storage,the risk of false generation of content and unfair competition,and criminal risk.Therefore,in the face of the reality of tension between data security and data utilization principles,adhere to and implement the principle of integrated security and development of data security and the principle of classification and grading of data utilization,adhere to the“personal-property”differentiated data classification model,and realize the“identification-traceability-exploitation”and“identification-traceability-exploitation”of virtual digital person data.The compliance path of“identification-traceability-regulation”is realized,so as to clarify the scope and mode of data collection and the regulatory boundaries of data security risks between enterprises and the government.