My first podcast / field trip with my colleagues at Cleantechnica. (Screenshot from Cleantechnica podcast)

China’s Most Updated Autonomous Driving Framework Makes Both Carmakers & Operators (Owners Included) Liable in a Crash


Support CleanTechnica's work through a Substack subscription or on Stripe.

In this week’s CleanTechnica YouTube gathering, fellow CleanTechnica writers Zach Shahan, Larry Evans, José Pontes, and I over-extended the allotted time of 30 minutes and talked for an hour and fifteen. The discussions were continuous and spontaneous, layered, one topic opening up another. One of the points we raised and discussed was autonomous driving laws in China, compared to what is going on in the US and Europe.

First of all, I want to clear what I said about liability. I used the word “driver” loosely. In the podcast I said that in China, if a road crash happens on a Level 5 vehicle, the liability goes to the owner (a refined translation of the law), which specified the “operator” as the owner. This is what I meant — the “driver” is the owner in private vehicles and the operator for commercial transportation.

China’s edge in autonomous driving

China is the world’s largest testing ground for autonomous vehicles, but until recently, legal ambiguity clouded the question of who should be held responsible when a self-driving car causes a crash. With the rollout of new municipal regulations in Beijing and Shanghai — and reinforced by national-level guidelines — China has now put into place the clearest liability system anywhere in the world for Level-4 and Level-5 autonomous driving.

The result is a decisive shift away from blaming the human occupant and toward assigning fault to the companies that design, manage, and operate self-driving fleets and owners of private vehicles.

Beijing’s Autonomous Vehicle Regulation, which took effect on April 1, 2025, establishes that the manufacturer of an autonomous vehicle “shall assume the main responsibility for vehicle quality and consistency of production.” The regulation requires carmakers to guarantee functional safety, ensure the reliability of autonomous-driving systems, and strictly manage software updates. (Note a documentation update was released this December but contained basically the same items and only the way the content’s subsections changed.)

At the same time, companies operating autonomous vehicles — whether robotaxi fleets, autonomous freight services, or other L4 and L5 deployments — must take on operational liability. Beijing requires operators to implement formal safety-production systems, continuously monitor vehicle status and surroundings, and report data to city authorities. The law’s language is explicit: operators must “perform the principal responsibility for safe production,” while manufacturers must “undertake the primary responsibility for product quality.” In a crash investigation, these entities become the starting point for liability analysis, not the occupant.

Differences in the megapolises

Shanghai’s regulatory framework, which governs the testing and commercial operation of intelligent connected vehicles, reinforces this shift. Under its measures for intelligent-vehicle testing and application, Shanghai requires any company deploying autonomous vehicles to hold the appropriate road-transport operator qualifications and to maintain a complete safety-production management system. During operation, such companies must inform users of residual risks, take necessary safety precautions, and comply with monitoring and reporting requirements. Most importantly, Shanghai explicitly states that if a self-driving vehicle operating in autonomous mode causes a traffic accident, the entity conducting the autonomous-vehicle operation “shall first bear the corresponding compensation liability.” Only after compensating the victim may the operator pursue recovery from the manufacturer or other responsible parties. This establishes a clear liability chain that begins with the operator — effectively mirroring legal expectations placed on public-transport and logistics companies.

These regulatory structures matter most in real-world scenarios. Consider a Level-4 autonomous vehicle cruising on a highway while the occupant watches a movie or sends messages. If the vehicle strikes a pedestrian or collides with another car, China’s laws do not treat the occupant as a negligent driver. When a vehicle is operating in high-level autonomous mode, human intervention is neither required nor expected; therefore, legal responsibility shifts away from the human inside the cabin. Instead, the operator’s safety systems, maintenance, and monitoring practices — and the manufacturer’s technical reliability — become the focal points. Investigators look to software behavior, sensor performance, operational protocols, and system logs to determine fault. The premise is simple: a passenger in an L4 or L5 vehicle is truly a passenger.

On the topic on insurance

Zach made a comment on insurance and how it will apply to other markets like Europe. In China, here is how it goes:

Beijing requires both manufacturers and operators to carry mandatory traffic-accident liability insurance, carrier liability insurance, and additional commercial coverage. Shanghai imposes similar insurance obligations to ensure that victims of autonomous-vehicle crashes can be compensated promptly, without legal uncertainty. This insurance requirement essentially embeds consumer protection into the operational framework. Instead of waiting for courts to determine whether a human or machine was responsible, victims can be compensated immediately, and insurers can later pursue recovery claims from responsible companies based on technical evidence.

China also mandates comprehensive data recording for all autonomous-vehicle tests and commercial operations, including continuous monitoring of vehicle state and environmental conditions. National guidance requires accident-data retention from at least 60 seconds before impact to 30 seconds after. These “black-box”-style requirements are designed to create a reliable evidentiary trail for crash investigations, allowing regulators and insurers to reconstruct events accurately and assign liability to the correct entity — typically the operator or manufacturer.

A deeper dive on private vehicles

Importantly, this framework applies not only to commercial robotaxis and fleet vehicles but also to private vehicles equipped with L4 and L5 autonomous-driving systems. While private cars with driver-assist features (L1/L2 or conditional L3) remain under the traditional rule that the human occupant is legally responsible, a private car operating at true L4 or L5 autonomy is treated similarly to a fleet-operated vehicle. In these cases, the occupant is legally a passenger and bears no driving responsibility.

Liability instead flows to the manufacturer if the crash was caused by system defects, or to the owner/operator if the crash resulted from negligence, misuse, or failure to maintain the vehicle and its autonomous systems. If the owner also functions as the operator, they may bear operational liability, but only in cases of improper handling or failure to meet legal operational requirements. Otherwise, responsibility resides with the system manufacturer or the licensed operator managing autonomous mode. Insurance frameworks similarly attach to the vehicle and its autonomous system rather than the occupant, ensuring coverage regardless of whether the car is privately or commercially owned.

China’s clear policy is an example for the rest of the world

China’s liability framework is different from those in the United States and Europe.

By establishing a uniform principle — that responsibility rests primarily with the companies designing and operating autonomous systems, even for private vehicles — China has created a foundation for scaling up robotaxis, driverless delivery, and autonomous freight at a national level. This same principles (not the laws) will ultimately affect flying cars and flying taxis and have become a template for operations of electric flying vehicles to be eventually deployed in the Middle East.

These policies (be these national or provincial) also ensure that private owners of highly automated vehicles can operate them legally while providing clear guidance for accident investigations and compensation.

China’s autonomous-driving regulations mark a decisive legal shift: when a Level-4 or Level-5 vehicle is operating in autonomous mode, the human occupant is not considered the driver.

Liability for crashes falls on the company operating the service and, where relevant, the manufacturer responsible for system performance and software integrity. By making this principle explicit — and by backing it with mandatory insurance, data requirements, and operator-licensing obligations — China has created the world’s most comprehensive and predictable liability structure for the coming era of driverless mobility.


Sign up for CleanTechnica's Weekly Substack for Zach and Scott's in-depth analyses and high level summaries, sign up for our daily newsletter, and follow us on Google News!
Advertisement
 
Have a tip for CleanTechnica? Want to advertise? Want to suggest a guest for our CleanTech Talk podcast? Contact us here.
Sign up for our daily newsletter for 15 new cleantech stories a day. Or sign up for our weekly one on top stories of the week if daily is too frequent.

CleanTechnica uses affiliate links. See our policy here.

CleanTechnica's Comment Policy


Raymond Tribdino

Raymond Gregory Tribdino, or Tribs, is an automotive and tech journalist for over two decades, a former car industry executive, and professor with deep roots in the EV space. He was an early contributor to EVWorld.com (1997-1999), was the motoring and technology editor for Malaya Business Insight (www.malaya.com.ph) and now serves as Science and Technology Editor for The Manila Times (www.manilatimes.net), along with co-hosting "TechSabado" and "Today is Tuesday." He's passionate about electrification, even electrifying his own motocross bike.

Raymond Tribdino has 327 posts and counting. See all posts by Raymond Tribdino