As the application areas of AI technology continue to expand，autonomous driving technology is having a profound impact on economic and social systems. In the process of commercialisation of autonomous driving technology，issues such as data privacy，data security，non-interpretability of algorithms，and algorithmic discrimination are becoming the focus of attention and difficulties in technology governance. Due to the dependence on deep learning algorithms，the field of self-driving cars presents characteristics such as diversification of responsibility subjects and blurring of causal chains，which leads to the fact that ethical principles and governance means applicable to traditional technologies are difficult to address the special risks in the field of self-driving cars. Based on theoretical research and industrial policy practice，this paper argues that the basic ways to solve the governance challenges of autonomous driving are as follows：common governance betweens takeholders，risk sharing based on residual risks，and progressive governance that synergises governance rules and governance practices.