China is moving autonomous driving closer to the center of its national traffic-law framework. On August 25, China’s top legislative body began considering proposed revisions to the country’s road-safety statute, including a new section for autonomous vehicles. The proposal does not authorize a nationwide rollout of driverless vehicles overnight. It does something more foundational: it begins to define how autonomous systems fit into rules originally written for human drivers.
Xinhua said the proposal covers requirements for public-road operation, procedures when a traffic rule is breached, and the insurance issues that follow. Reuters added that the proposal distinguishes autonomous-driving functions from driver-assistance functions and could make manufacturers or importers responsible for violations while a vehicle is operating in autonomous-driving mode.
The proposal is still in its first review. It is not enacted law, and the available reporting does not establish a final set of rules for every automation level or use case. But the decision to create a specific chapter is significant. It recognizes that an AI-driven vehicle cannot be governed cleanly by rules that assume a human is always making the immediate driving decisions.
Separating Automation From Driver Assistance
The distinction between autonomous driving and driver assistance may sound technical, but it is central to legal accountability. Many vehicles already have systems that can help with lane keeping, braking, parking, or highway driving. In those cases, a human remains responsible for supervising the vehicle and taking over when necessary. A genuinely autonomous function raises a different question: if the system is in control, who answers when it breaks a rule or causes harm?
Reuters’ account of the draft points toward an answer. When a vehicle is operating in autonomous-driving mode, the manufacturer or importer could be responsible for traffic violations. When a person is driving manually or using an assisted-driving feature, existing rules would still apply. That approach is not a complete liability code, but it begins to allocate responsibility based on the system’s operating mode rather than treating all software-enabled driving as the same.
This is one reason why precise definitions matter. A carmaker may market a feature as advanced, while a consumer may assume it is more capable than it is. A national law that distinguishes the categories can create a clearer baseline for regulators, insurers, police, manufacturers, and drivers. It can also discourage companies from using ambiguous language about what a vehicle can do.
China has already become a major testing ground for autonomous mobility. Baidu, Pony.ai, WeRide, XPeng, and other companies have pursued robotaxi services, advanced driver-assistance systems, and freight or delivery applications. EastFrontier’s recent report on Pony.ai’s second-quarter robotaxi revenue growth showed why commercialization has become as important as demonstrations. Companies need a route from local pilots to services that can operate within clear and repeatable rules.
From Local Pilots to National Legal Infrastructure
Until now, much of China’s autonomous-driving development has been shaped by city-level testing permits, regional regulations, and controlled commercial zones. That has allowed local governments to encourage innovation while retaining oversight. It has also produced a patchwork. A national traffic-law amendment could create common definitions and legal principles that apply beyond an individual pilot area.
That does not mean local experimentation becomes irrelevant. Cities will still have different road conditions, traffic density, weather, public-transit patterns, and industrial priorities. But a national framework can reduce uncertainty about the basic legal status of autonomous operations. It can also make it easier for companies to design vehicles and services for more than one jurisdiction.
Insurance is another difficult piece. The draft’s inclusion of insurance arrangements signals that lawmakers understand autonomous driving is not only a technology question. When a collision occurs, insurers need to know what data can establish the system’s mode, whether a human was expected to intervene, and who bears the initial cost while liability is investigated. The sources reviewed do not provide a final insurance model, so it would be premature to say the matter is settled. The important point is that insurance is being placed inside the legal discussion from the start.
China’s companies are already thinking globally. EastFrontier has covered Pony.ai and Uber’s plan for 2,000 robotaxis in Europe, demonstrating that Chinese autonomous-driving firms must meet rules across different markets. A clearer domestic framework will not replace foreign regulation, but it can give companies a more structured home base from which to build technology, safety practices, and commercial operations.
What the Draft Still Needs to Resolve
The current proposal marks a beginning, not an endpoint. The public reporting leaves open many detailed questions. How will officials verify that autonomous mode was active? What data must a vehicle retain after an incident? How will responsibility be divided among a vehicle manufacturer, a software provider, an owner, an operator, and a remote support team? What happens when a human is asked to take over but cannot do so in time?
These questions are not unique to China. Every country trying to regulate autonomous vehicles faces them. China’s advantage is the scale of its testing ecosystem and the depth of its automotive and AI industries. Its challenge is to turn those advantages into rules that are clear enough to protect road users without freezing technology at an early stage.
The draft’s focus on definitions, operating conditions, violations, and insurance is a sensible place to start. It moves the debate away from abstract claims that autonomous driving is either inevitable or dangerous and toward the practical systems needed to govern it. A vehicle may be capable of navigating a road, but society still needs to decide what happens when the machine is wrong.
For China’s AI and mobility companies, the proposal is both an opportunity and a warning. A dedicated legal chapter could help create a path to wider deployment. It also makes clear that deployment will come with defined responsibilities. The next stage of autonomous driving in China will not be decided only by sensors, models, and road tests. It will be shaped by the rules that determine who is accountable when those systems meet the real world.