China proposes tougher car safety checks, 30,000 km tests and limits on novelty designs

A year of testing before the next new model

China is proposing a minimum of 30,000 kilometres of vehicle reliability testing and at least one year of environmental validation as part of a broader effort to stop manufacturers putting new designs into production before their safety has been adequately checked. The proposals would also prohibit fully concealed door handles and folding or flexible displays, while preventing the use of zero gravity seats, rotating seats and seat arrangements that turn into beds while a vehicle is moving.

The Ministry of Industry and Information Technology, the Ministry of Public Security, the Ministry of Ecology and Environment and the State Administration for Market Regulation released the consultation draft on October 10. Its title translates as the Notice on Further Strengthening Management of Automotive Product Innovative Design and Research, Development, Testing and Validation.

The proposed enforcement dates are January 1 and July 1, 2027. From the first date, products with missing, insufficient or falsified testing evidence would be refused entry in the ministry's vehicle product announcement, a necessary step for market access. Models already listed would have until July 1 to submit supplementary documentation. Products that miss that deadline and present safety hazards would face production suspension and recall.

These are proposed requirements, not a final notice already in force. Their scope nevertheless reaches beyond a list of prohibited features. The draft would require manufacturers to demonstrate how a vehicle was validated, disclose the limits of innovative functions and maintain consistency between tested components and the vehicles delivered to customers.

Manufacturers have already begun internal reviews. Several company representatives said they organised study sessions over the weekend following publication and planned to circulate the requirements throughout their businesses. Some also said local authorities were organising implementation preparations.

What the 30,000 kilometre requirement would test

The draft separates several tasks that are sometimes grouped together under the word testing: complete vehicle reliability and durability, structural reliability, software function reliability and environmental adaptability. Passing a final inspection would not substitute for evidence across these areas.

Whole vehicle reliability trials would cover rough roads, ordinary roads, highways and mountain roads, with a minimum distance of 30,000 kilometres. Structural checks would examine fatigue life and mechanical performance under extreme conditions for the body, frame, suspension, steering and braking components. Fatigue means damage that builds through repeated loading, even when an individual load does not immediately break a part.

Wiring harnesses and exhaust treatment equipment would also fall within checks addressing safety, environmental performance and vehicle reliability. The approach connects component performance with the conditions in which the complete vehicle operates, rather than treating each successful laboratory result as sufficient on its own.

A parallel standards proposal from the electric vehicle subcommittee of China's national automotive standardisation committee would increase the reliability driving distance for battery electric and fuel cell vehicles from 15,000 to 30,000 kilometres. That is a doubling of the existing distance and would bring it into line with the figure used for conventional fuel vehicles. This standards amendment is also a proposal, distinct from the four departments' draft notice.

Two winters and one summer, with conditions on substitutes

Environmental validation would last at least one year and include two winters and one summer. Manufacturers would have to use multiple test vehicles and cover all key configurations, rather than assume that a result from one version necessarily applies to every variant.

The draft allows testing in locations experiencing the opposite season and simulations in indoor environmental chambers. Those alternatives would have to provide validation at least equivalent to actual road testing. Where interacting environmental factors cannot be reproduced adequately, real road trials would still be required.

This distinction matters because permission to use simulation is not permission to replace every physical test. A chamber can reproduce selected conditions, but the draft specifically preserves road testing for combinations of conditions that cannot be represented with equivalent effectiveness. The one year minimum and seasonal coverage requirements would therefore operate together, not as interchangeable options.

Which cabin features would change?

Fully concealed door handles would be prohibited under the proposal. Innovative handle designs would need checks of their function in extreme environments and their usability in emergencies, along with mechanical redundancy and enough space for a hand to operate them. Mechanical redundancy means retaining a way to release the door without relying solely on an electrical command.

The draft would also require powered door closing systems to prevent trapping occupants. Zero gravity seats would need both protection against trapping and collision safety validation, but could not be used in that position while driving. Rotating seats and functions that flatten seats into beds would likewise be unavailable during vehicle movement.

Central entertainment screens would require collision safety assessments, and folding or flexible displays would not be permitted. Unusually shaped steering wheels are also among the designs whose use and validation the proposal addresses. The common requirement is that a feature's appearance or convenience cannot stand in for evidence that it is safe in its intended setting.

The restrictions are not identical for every feature. Some designs would be prohibited outright, while others could remain if manufacturers validate them and limit when they operate. In particular, the proposed restriction on reclining or rotating seating concerns use while the car is moving, rather than a blanket ban on every such seat.

Emergency controls must remain physical

Controls for emergency escape, emergency braking and cutting electrical power would need physical operating mechanisms and conspicuous labels. This would preserve a directly identifiable means of action when a driver or passenger has little time to respond.

For functions such as lights and windscreen wipers, a manufacturer using virtual controls would have to assess operating reliability across different situations and retain physical controls. A working touchscreen interface in ordinary conditions would therefore not, by itself, establish compliance with the proposed requirements.

The draft also addresses who can issue commands. While a vehicle is moving, people other than the driver would not be allowed to directly operate key driving safety functions through voice or gesture interaction. The functions listed include defrosting and demisting, lights, wipers, doors and the front bonnet.

These provisions distinguish between adding another interface and removing an established safety control. Voice commands, gestures and screen menus can introduce new ways to operate a vehicle, but the proposal would restrict their authority where unintended commands could interfere with driving or emergency action.

Software updates would need evidence too

Software reliability validation would run through development, including system testing and checks in the complete vehicle. Updates delivered remotely, commonly called OTA updates, would also need sufficient testing. Filing materials would include the manufacturer's internal approval records.

Li Yanwei, a member of the expert committee of the China Automobile Dealers Association, described the proposal as moving regulatory attention from whether a finished product meets requirements to whether development has been adequately validated. He said the scope for selling a vehicle first and repairing problems through later software updates would narrow.

Discussing software changes, Li said, in translation:

Software upgrades also need sufficient validation, and internal approval records must be retained.

The proposal would add 51 vehicle technical parameters to market access applications: 41 relating to innovative designs and 10 relating to development testing and validation. Manufacturers would fill them in according to the product's actual configuration and submit a commitment that the information is complete, accurate and consistent.

From January 1, 2027, newly submitted models incorporating innovative designs would need the relevant parameters and supporting validation materials. Checking those records before a model is listed would be followed by continued scrutiny of production consistency after manufacturing begins. The regulatory chain would therefore extend from development evidence to the performance of production vehicles.

Advertising would have to explain functional limits

The draft would bring marketing conduct and repair economics into vehicle market access management. Its advertising provisions require manufacturers to describe innovative features objectively, including their operating boundaries, restrictions and development validation.

False, exaggerated or misleading promotion would be prohibited. Manufacturers would also be barred from using scenarios that violate road traffic rules to promote a function or encouraging drivers to neglect their responsibility for safe driving.

These requirements connect the engineering definition of a feature with the customer's understanding of it. A system that works only under specified conditions should not be advertised in a way that suggests unrestricted operation. The proposed limits on seats during driving and on passenger voice commands are examples of boundaries that would need to be reflected accurately in promotion.

Repair economics is identified as another new element of market access management, but the reporting does not establish a detailed cost threshold or assessment method. It should not be read as a confirmed cap on repair bills.

A broken brake bracket shows why validation matters

The debate has been sharpened by a braking test involving three Zunjie V800 vehicles. During repeated emergency stops from 100 km/h, their brake pedal brackets reportedly fractured on the third, fourth and second full braking attempts respectively. The repeated failure across three vehicles made the strength of the assembly, its validation conditions and its production quality central issues.

Zunjie described the exercise as a nonstandard extreme test. The company said its braking system had been developed and validated against requirements above relevant national and industry standards, and that it had received no reports of equivalent failures in actual customer use. It promised to improve the component design and provide free upgrades to customers who had already taken delivery.

Those statements do not settle the cause or establish the extent of any defect. Possible explanations discussed in connection with the failures include variation in manufacturing strength and insufficient reassessment of an existing bracket when used in a different installation. Neither has been established as the cause.

The distinction between pedal force and bracket loading is also relevant. The force applied by a driver's foot is not necessarily the force experienced at a particular support or fracture point. The pedal's geometry, pivot position, installation and direction of loading can change the stresses within the assembly.

A documented earlier case illustrates the risk associated with material changes. In 2020, Ford recalled 2,627 imported Mustang vehicles in China after changing brake pedal bracket material from nylon to polypropylene. The recall explanation identified insufficient allowance in engineering specifications and design validation for the resulting reduction in peak load strength, with a risk of fracture during emergency braking. That case does not prove the cause of the V800 failures, but shows why material choice must be assessed together with design and validation.

China's Regulations on the Administration of Recall of Defective Auto Products also distinguish a defect from a simple failure to meet a numerical standard. Article 3 includes other unreasonable dangers to personal or property safety within the definition of a defect. Article 12 requires a manufacturer that learns of a possible defect to investigate and analyse it, and to recall the product if a defect is confirmed. A promised free upgrade does not by itself resolve that legal assessment.

The timetable and the questions still open

The proposal comes amid a wider series of vehicle safety measures. The reported timetable separates standards already issued or implemented from the future dates proposed for this notice:

• January 28, 2026: the mandatory door handle safety standard, GB 48001 (2026), was approved and published, addressing mechanical release after power loss and visible interior handle safety markings.
• June 2026: GB 47955 (2026), covering combined driving assistance systems, was approved and published, with implementation scheduled for January 1, 2027.
• July 2026: GB 44721 (2026), covering Level 3 and Level 4 automated driving systems, was published, with implementation scheduled for July 1, 2027.
• From late July: stricter application reviews barred newly submitted passenger vehicles from factory installation of blue driving assistance signal lamps because blue did not comply with existing exterior lighting colour standards. Existing vehicles were not subject to compulsory action at that stage.
• October 1: GB/T 47315, the road simulation bench test method for vehicle structural durability, took effect.
• October 10: the four departments released the innovation design and development validation consultation draft.
• January 1 and July 1, 2027: the draft proposes new application requirements and the deadline for supplementary evidence from products already listed.

The driving assistance standard covers Level 2 systems, where the driver remains responsible for driving, and sets requirements for interfaces, driver monitoring and data recording. The separate automated driving standard addresses Level 3 and Level 4 systems. These are distinct regulatory tracks, not interchangeable descriptions of the same technology.

The China Association of Automobile Manufacturers publicly supported the proposal and issued an industry appeal on October 11. It urged companies to put safety assessment at the start of product innovation, strengthen complete vehicle reliability and complex environmental testing, and support implementation through better development processes.

The pressure on development schedules is substantial. A cited McKinsey comparison puts current domestic new vehicle development at about 24 months, against roughly 60 months in the conventional fuel vehicle era. That is 36 months shorter, or a 60% reduction. It is a comparison of development periods, not proof that every faster programme lacks adequate testing.

Li identified unresolved details, including the definition of extreme conditions, loading standards, failure criteria, climate coverage and validation intensity. He also argued that a road simulation bench method cannot replace risk identification and acceptance criteria for each component. No consultation closing date or confirmed final wording is established here. The proposed minimum distances and periods therefore need to be distinguished from the detailed rules that would determine whether a test programme is sufficient.

Key Points

• Four Chinese departments have proposed stronger checks of vehicle innovation, development testing and production consistency.
• Reliability trials would cover at least 30,000 kilometres, with environmental validation lasting at least one year and including two winters and one summer.
• Fully concealed handles and folding or flexible screens would be prohibited; certain seating functions would be restricted during driving.
• Emergency controls would remain physical, while software updates would require testing and internal approval records.
• The draft proposes January 1, 2027 requirements for new applications and a July 1 deadline for supplementary evidence from existing listed products.
• The V800 brake bracket failures remain subject to unresolved questions about cause, risk scope and the effectiveness of promised improvements.
• The notice is still a consultation draft, and detailed test criteria and final wording have not been confirmed.