With the rapid development of artificial intelligence, accidents caused by AI products and services are increasing day by day. When accidents cause personal or property damage but the responsible party cannot be determined, artificial intelligence insurance can balance victim relief and intelligent technology innovation, promoting the sustainable development of artificial intelligence. Accelerate the legal construction of artificial intelligence insurance, and highlight its practical significance in the specific application fields represented by autonomous vehicle. In the field of autonomous vehicle, there is no insurance system specially designed for the characteristics of artificial intelligence products and services in China at present, and the traditional compulsory traffic insurance is still the main insurance. Compulsory traffic insurance has the attribute of no fault insurance. Its internal logic is still that the driver or manager is the first person responsible for motor vehicle traffic accidents, and the damage caused by autonomous vehicle is likely to originate from artificial intelligence products and services themselves. With the iterative upgrading of AI products and services, coupled with the "black box" nature of AI products, it is sometimes difficult to determine the cause of damage caused by autonomous vehicle, and it is difficult to identify the subject of responsibility. In order to protect the rights and interests of victims, it is particularly necessary to implement new artificial intelligence insurance in the field of autonomous vehicle. China has the foundation for the legalization of artificial intelligence insurance in the field of autonomous vehicle. At present, China's insurance law, the Regulations on Compulsory Automobile Traffic Accident Liability Insurance and other laws and regulations are the basis for promoting the legalization of artificial intelligence insurance in the field of autonomous vehicle. Some parts of our country have started exploring the legalization of artificial intelligence insurance. As stipulated in Article 56 of the Shenzhen Special Economic Zone's Regulations on the Promotion of the Artificial Intelligence Industry, it is necessary to explore and improve a specialized insurance compensation system that adapts to artificial intelligence products and services, and provide full chain insurance protection for artificial intelligence products and services. The Regulations on the Management of Intelligent Connected Vehicles in Shenzhen Special Economic Zone encourage insurance companies to develop insurance products that cover the full chain risks of design, manufacturing, use, operation, data and algorithm services. Considering the particularity of artificial intelligence products and services and promoting the legalization of artificial intelligence insurance in the field of autonomous vehicle, the following issues need to be considered. Firstly, in the design of insurance types, compulsory liability insurance should be taken as the basis. Artificial intelligence mandatory insurance should not only cover human behavior and fault, but also comprehensively consider various potential responsibilities of artificial intelligence. Second, in terms of accountability for damage caused by products and services, it is necessary to gradually transition to focusing on the producers of products and service providers. Whether the AI insurance adopts the product liability insurance model or the tort liability insurance model, the producers of AI products and service providers should bear the main responsibility, thus forcing the producers of AI products in the field of autonomous vehicle to constantly improve their production technology and the service providers to constantly optimize their service capabilities, and promoting the development of the AI industry. Thirdly, in the formation of insurance products, priority can be given to using the model of artificial intelligence product producers and service providers to purchase insurance products, as well as the model of artificial intelligence product users purchasing insurance products, and the model of artificial intelligence enterprises cooperating with insurance companies to develop insurance products. The advantage of the latter is that both can grasp more abundant and effective data on the use of artificial intelligence products and services, and continuously improve the design of artificial intelligence insurance products. Fourthly, in the design of supporting insurance products, it is recommended to establish a risk compensation fund. Artificial intelligence products and services may have a risk of large-scale infringement, and potential responsible parties of the artificial intelligence system or beneficiaries of the artificial intelligence industry chain should jointly establish a risk compensation fund to compensate for damages not covered by compulsory insurance. Fifthly, in the design of insurance product support, it is recommended to connect with the existing social assistance system and establish a social assistance fund specific to artificial intelligence products to effectively respond to the relief of victims. Sixth, in terms of personal privacy and data security, the implementation of artificial intelligence insurance should fully safeguard personal privacy and behavioral data security, and avoid the risk of data leakage or abuse during the use of data. (Liao Xinshe) (Authors Su Xiangju and Liu Wei, affiliated with Tsinghua University Nuclear Research Institute and Civil Affairs Vocational University)