【 Zizhengchang 】 Author: Li Dejia (Associate Professor at Beijing Normal University Law School) Recently, two drivers were drunk and turned on the "intelligent assisted driving" function of their vehicles in the Yunwu Mountain Tunnel of Chongqing Chengyu Ring Expressway. They then fell asleep in the tunnel, causing the vehicle to come to a stop. The two have been investigated for the crime of dangerous driving and will face criminal punishment. With the rapid development of artificial intelligence technology, intelligent driving cars have gradually entered the market, entering thousands of households and bringing more technological experiences to consumers. At the same time, cases of using "intelligent assisted driving" as a "proxy driver" under the influence of alcohol have occurred from time to time. From the perspective of individual case handling, the current criminal law provisions on the criminalization of drunk driving are sufficient to cover such behavior, but how to further refine the technical level, operational scenarios, and responsibility allocation on the basis of existing rules has become an unavoidable legal issue for the healthy development of autonomous driving in the future. The regulatory rules need to be further classified and refined At present, China has gradually established a hierarchical management system for autonomous vehicle. Based on the national standard Classification of Driving Automation, it is divided into levels 0 to 5, corresponding to six levels of "emergency assistance, partial driving assistance, combined driving assistance, conditional automatic driving, highly automatic driving, and fully automatic driving". The core difference between levels lies in the degree to which the driving automation system can perform dynamic driving tasks, role allocation, and whether there are design operating range limitations. During the road testing phase, regulatory authorities have put forward clear requirements for the participants, vehicle conditions, and testing sites, emphasizing the need to conduct testing in a controllable environment and gradually verify technical safety. In the admission and on road stage, "autonomous driving" will be limited to levels 3 and 4 of the driving automation level in the "Classification of Automotive Driving Automation", and management experience will be accumulated through pilot projects. For the so-called "intelligent assisted driving" that has been widely applied in the market, regulatory authorities have repeatedly reiterated that its nature is only "combined driving assistance" rather than "autonomous driving". They explicitly require that misleading propaganda such as "autonomous driving" should not be used, and emphasize that drivers must monitor road conditions throughout the process and be ready to take over at any time. The recent multiple "drunk driving cases" involving only the widely used Level 2 combination driving assistance technology. In the context of this system, in the relevant "Intelligent Driving Drunk Driving Case", the court clearly pointed out that the level 2 combination driving assistance function does not change the legal obligation of drivers to continuously observe and take over at any time. Drunk driving constitutes the crime of dangerous driving, which fully clarifies the application boundary of existing rules in the context of new technology. In the future, the institutional rules that need to be further refined mainly lie in three aspects: firstly, there is a lack of more unified and specific normative expressions for the applicable boundaries and management requirements of different levels of systems in different operating scenarios. Secondly, relevant regulations are scattered across multiple departmental documents and local pilot projects, and there is still a need for further connection between higher-level legal norms, departmental regulations, and technical standards. Thirdly, for the already popular driving assistance functions, there is still a lack of a set of specific rules that can be directly implemented and referenced by law enforcement and the judiciary, including unified risk notification before implementation, mandatory prompts and takeover mechanisms during use, and allocation of driver and producer responsibilities based on data records afterwards. To address the challenges of technological iteration through graded responsibility legislation, autonomous driving will not completely replace human driving in the short term, but will be in a long-term state of interweaving and coexisting with the two. How to reserve institutional space in advance for the maturity of higher-level autonomous driving technology in the future through graded responsibility is a problem that needs to be considered at present. It is necessary for future autonomous driving legislation to attempt to establish a graded responsibility system that matches the level of technology on the existing basis. Firstly, Level 1 partial driving assistance and Level 2 combined driving assistance adhere to the principle of "human responsibility", taking the driver's obligation to continuously observe and take over at any time, as well as the "zero tolerance" for drunk driving, speeding and running red lights, as an unshakable responsibility baseline. Enterprises only bear supplementary responsibility for serious explanations and reminders of defects. Secondly, Level 3 conditional autonomous driving and Level 4 highly automated driving implement a "strict liability+fault liability" composite model: accidents caused by system defects, hardware failures, and data security issues are strictly borne by the producers and system providers, and costs are allocated internally through insurance and recovery; If the driver refuses to take over under clear and feasible takeover instructions, they still bear the responsibility for their negligence and rely on standardized data recording devices to clarify the boundary of human-machine responsibility. Thirdly, a responsibility framework centered around manufacturers, system operators, and data controllers should be reserved for Level 5 fully autonomous driving, supplemented by mandatory insurance, defect recalls, and continuous safety operation records. Through such a hierarchical responsibility system, dynamic matching is achieved based on the principle of "who controls, who is responsible", which not only avoids the vacuum of no one being responsible, but also prevents the simple accumulation of new risks on individual drivers. Building a diversified and collaborative governance pattern for autonomous driving. The future development of technology and institutional construction is a process of mutual influence and shaping, which must be synchronously promoted, continuously adjusted, and adapted through continuous interaction. How to develop autonomous driving technology, how to define functional boundaries, and how to enter road operation ultimately need to be confirmed and guided by rules. Guided by the rule of law, it is necessary to incorporate the development of autonomous driving into a predictable and constrained regulatory track through timely and accurate institutional supply in the process of continuous technological iteration. Strengthen the coordination between top-level design and standard system. On the one hand, it is necessary to clarify the legal status of autonomous vehicles and the basic rights and obligations of relevant parties at the legal level, providing a unified normative basis for departmental regulations and technical standards; On the other hand, it is necessary to strengthen the coordination between legislation in areas such as road traffic safety, product quality, network and data security. Weave a regulatory and accountability network that runs through the entire process from a legislative perspective. Reduce risks from the source by specifying market access conditions and safety assessment requirements for autonomous vehicles beforehand; In the process, clarify the continuous monitoring obligations of the operating entity and the intervention measures of regulatory authorities to prevent operational risks; Afterwards, improve the systems of data retention, accident investigation, defect recall, etc., to form a closed loop between accountability and rectification implementation. Utilize the exploration and feedback functions of local pilot projects. Within the scope of legal authorization, support pilot areas to carry out controllable explorations around the division of operational scenarios, refinement of responsibility rules, standardization of data governance, and accident investigation mechanisms. Establish an experience feedback channel from local pilot to national legislation, so that good experiences and practices verified through practice can be timely upgraded to higher-level unified norms. For the legal regulatory path of autonomous driving, the most important thing is to follow the trend of the development of autonomous driving technology, and to accelerate the improvement of the rule system, responsibility structure, and regulatory mechanism. Seize the critical window period of accelerated development of autonomous driving technology, coordinate safety and development with the rule of law as the main line, continuously refine operational scenarios, clarify responsibility boundaries, and improve governance mechanisms within the legal framework, in order to achieve refined governance of autonomous driving and ensure safe, controllable, and sustainable development of technology on the track of the rule of law. (New Society)