As a typical representative of generative artificial intelligence, the algorithm-driven virtual digital human can interact with humans through autonomous analysis of massive big data (text, pictures, audio, video, etc.) on the basis of metadata feeding. The virtual digital human equipped with ChatGPT technology will realize the 3D scene application of algorithm-driven + human modeling/virtual image. Its output content (artificial intelligence products) includes music, dance, text and other forms. Because of the four characteristics of interactivity, digitization, personification and uniqueness, the algorithm-driven virtual digital human has produced controversial issues such as the protection of the rights of the virtual digital human itself and the demarcation of the ownership of the relevant content generated due to its interactivity.The first category is the damage caused by the algorithm-driven virtual digital human to the subject of data rights in the process of pre-training data collection and optimization of training data at the beginning of research and development; the second type is the infringement that the algorithm-driven virtual digital person itself or the R & D subject who has property rights and interests in it may suffer. Specifically, it can be divided into two situations:the ownership dispute that may be faced by the products with original content independently generated by the algorithm-driven virtual digital person and the different ownership classification generated by the algorithm-driven virtual digital person based on different creation modes. In the case of unilateral creation of the operating company, the rights and interests of the algorithm-driven virtual digital person should be classified into the operating company. However, if the user is also involved in the creation process of the virtual digital person, it can be subdivided into three forms:‘materialism',‘creationism' and ‘cooperationism', which correspond to the three ownership subjects of platform ownership, user ownership and common ownership.At present, as a digital tool, virtual digital people do not have the personal dignity of the premise of personal interests, so it is not appropriate to identify them as having independent personality and enjoying personal interests. By deconstructing the various parts of the virtual digital person, such as portrait, name, voice and reputation, the interests generated mainly adopt the ‘property theory' to protect, while the reputation of the virtual digital person is usually relieved by the ‘same theory'. In addition, the copyrightability of intelligent virtual digital human products is essentially the copyrightability of artificial intelligence products based on deep learning technology, and there is a controversy in the academic community about whether it should be included in the public domain.In the process of the development of generative artificial intelligence, China can consider exploring the way of unified legislation of artificial intelligence. Specifically, the first is to balance the distribution of regulatory obligations between the generated artificial intelligence service provider and the government by enhancing the large-scale infrastructure service obligations of the generated artificial intelligence service provider. The second is to construct a moderate exemption rule for the metadata collection of the generated artificial intelligence service provider from the perspective of encouraging development. The third is to clarify that the artificial intelligence product should not enter the public domain directly at this stage, and its ownership division is mainly based on the agreement between the service provider and the user.