121 million vehicles "VS" 370 million vehicles.
The former is the cumulative number of vehicle recalls implemented in China by the end of 2025, while the latter is the national vehicle ownership by the end of June this year. This means that approximately one in every three vehicles has experienced a recall.
Most of these 121 million vehicles were accumulated during the era of internal combustion engine (ICE) vehicles, representing 22 years of accumulation. In 2004, China began implementing the defective automobile product recall system, which has been in use ever since. However, in recent years, as China enters the era of intelligent and electrified vehicles, Xiao Lingyun, Director of the Industrial Products Department of the Defective Product Recall Technical Center of the State Administration for Market Regulation (SAMR), pointed out at the 2026 TEDA Automotive Forum that the issues faced by recalls at this stage are more complex.
Vehicle recalls in the intelligent era should not be the sole responsibility of original equipment manufacturers (OEMs). The institutional standards for vehicle quality supervision and vehicle recalls are undergoing changes.
Quick-Built Cars: All Parties Share the Responsibility
In the current automotive market, a prominent issue affecting the consistency, stability, and safety of vehicle quality is the emergence of "quick-built cars".
A major manifestation of quick-built cars is the increasingly shorter development cycles for new models and the accelerating pace of market launches. In the past, developing a car took three to four years; nowadays, it must be completed in 18 months, 15 months, or even 12 months.
The compression of development cycles has quickly transmitted to the supply chain. Wang Xianbin, Partner and Vice President of the Research Institute at Gasgoo, stated at the 2026 AI-Defined Vehicle Forum hosted by Gasgoo that if a new car is developed within 12 months, the time left for suppliers to conduct development and validation might only be 3 to 4 months.
OEMs control product definition, procurement, and development milestones, making them a crucial party in compressing cycles and driving cost reduction, a fact that cannot be avoided. However, suppliers are also proactively adapting to this competitive logic. To secure orders from leading clients, some suppliers will voluntarily lower prices, commit to tight delivery milestones, and even compress their internal validation cycles.
For example, in the LiDAR industry, a couple of years ago, some suppliers lowered their quotes to levels unacceptable to other companies just to win orders. More than one supplier told Gasgoo that even if prices were squeezed, they had to grit their teeth and do it, "expanding scale first before considering profitability to avoid collapsing halfway through."
Zeng Qinglin, General Manager of the Yijing Auto brand, said at the conference that the essence of involution in industry competition is resource consumption in the same dimension and parameter gaming. Looking at the domestic auto market, there are many homogenized enterprises and products. Some companies lack innovative spirit and long-term planning; to seize market share, they have chosen the competitive approach of involution and price bidding.
Currently, the entire chain is running under this logic. Wang Xianbin stated, "After saving all this time and still failing to meet the milestones, the only option left is to cut into the validation process."
For instance, in winter testing, the effective testing window is usually only three to four months; if missed, one has to wait until the next year. Vehicles also need to be driven on real roads, and components must withstand tests of high and low temperatures, durability, and complex road conditions. These tests have clear time and scenario requirements; simply compressing the schedule makes it difficult to complete the corresponding validation work synchronously.
The validation work for new energy vehicles (NEVs) is even more complex. Batteries, intelligent configurations, and hardware-software coordination all need to be integrated into the development and testing processes. In the past few years, to meet market launch milestones, some automakers adopted downgraded solutions such as hardware pre-installation, launching the vehicles first and then supplementing some functions through over-the-air (OTA) remote upgrades. This practice can temporarily alleviate the pressure of launch schedules, but it also extends some development and validation work to after vehicle delivery.
Nowadays, there are fewer semi-finished cars, but practices such as white-box development and rapid integration of third-party solutions remain quite common in the industry. In addition, some novel styling designs in NEVs have also left safety hazards. Liu Zongcheng, President of Seres Group Technology Company, emphasized, "The biggest risk in the automotive industry is not being slow, but being quick-built." Ultimately, the price paid by problematic vehicles is recall, and "the recall cost could reach hundreds of millions of euros."
On August 21, nine automakers recalled approximately 4.276 million vehicles on the same day due to issues with the emergency mechanical release handles inside the cars. Hidden door handles have been a design selling point eagerly pursued by automakers in recent years, claiming to reduce air resistance and offer a clean appearance. However, in extreme accidents, the emergency mechanical release handles inside the car are close in color to the interior trim, making it difficult for occupants to find the door release handle, which instead hinders escape.
Advanced Driver Assistance Systems (ADAS) also face similar validation challenges. In 2024, multiple recalls initiated domestically due to ADAS defects involved 2.556 million vehicles, accounting for 20% of the total recalls for the year.
Xiao Lingyun emphasized at the conference that automakers must pursue genuine innovation, innovation that can truly be accepted by consumers.
Should Suppliers Also Be Held Responsible for Recalls?
Quick-built cars are precisely one of the problems exposed during the rapid development of China's intelligent and electrified automotive sector. The quality risks they bring are no longer solely the concern of OEMs; suppliers also play a role in this. Xiao Lingyun noted that vehicle recalls in the intelligent era have already encountered some new regulatory challenges.
For example, under the battery swap model, who is responsible for battery safety? ADAS are increasingly concentrated in a few specialized companies. If a defect originates from the ADAS, whether the system supplier can become the recall entity and what obligations they should bear also require a response from the regulatory system.
According to the current "Regulation on the Administration of Defective Automobile Product Recall", vehicle recalls mainly revolve around vehicle manufacturing enterprises. As the technical architecture and industrial division of labor for smart cars change, defects in batteries, software, ADAS, and other links increasingly require investigation and determination in combination with the technical responsibilities of different enterprises.
OEMs still need to be responsible for the final product, and the responsibilities that relevant suppliers should bear in the formation and rectification of defects also need to be further clarified.
In the era of ICE vehicles, the question of "who is the subject of vehicle recalls" was not difficult to answer. This is because OEMs bear the primary responsibility for the design, production, and validation of vehicles, while suppliers mainly provide components and technical support. Core technologies such as engines, transmissions, and chassis are either held within the OEMs themselves or provided by long-term cooperative suppliers.
The supply chain for smart cars, however, is much more complex. A single vehicle may integrate core components such as batteries, chips, software, and ADAS provided by multiple enterprises. In particular, intelligence has permeated the entire lifecycle of automotive manufacturing. Vehicle safety depends not only on the components themselves but also on the matching between different systems and the overall vehicle integration.
Moreover, the cooperation between OEMs and suppliers has already penetrated into product definition and joint development. For example, Seres has built a "1+2+N" strategic cooperation pattern: 1 is Seres itself, 2 are Huawei and CATL, and N represents numerous world-class partners. Yijing Auto is jointly created by Dongfeng and Huawei Qiankun. Zeng Qinglin summarized the cooperation between the two parties as "two-way joint definition, joint development, and joint office work."
This cooperation model allows different enterprises to jointly participate in vehicle development, which also requires defect investigations to consider more links. The problem might lie in a specific component, or it could stem from the matching between hardware and software, or anomalies that occur when different systems operate collaboratively. Merely inspecting the production links of OEMs may not quickly identify the source of the defect. Simply attributing the problem to component defects from suppliers might equally overlook issues in overall vehicle integration and system matching.
Software and data add new complexity. Traditional vehicle recalls mostly target hardware defects, whereas intelligent connected vehicles may encounter safety issues due to software upgrades, control strategies, or data processing.
Issues triggered by OTA upgrades have already attracted regulatory attention. Previously, certain models experienced failure of the automatic emergency braking function due to improper upgrades of image processing software. There were also models that, two years after an OTA upgrade, discovered that the upgrade had triggered new faults.
Such cases indicate that software upgrades themselves can also alter the functional performance of vehicles. Regulators need to grasp the upgrade content, the scope of impact, and the defect situation to determine whether recall measures are necessary.
Battery "power locking" is another example. Some automakers, under the name of "safety optimization," modify battery management system parameters without users' knowledge.
Zhang Xiaoyu, Executive Vice President of Changan Automobile, stated that many vehicles are equipped with driver assistance systems, but the actual usage rate by users is not high. The core reason is insufficient safety and the difficulty in building user trust.
Therefore, what the recall system needs to address is not only whether suppliers can become recall entities, but also how software defects are determined, how OTA upgrades are incorporated into regulation, and how OEMs and suppliers can jointly cooperate in defect investigation and rectification.
Regarding OTA recalls, relevant departments have already required enterprises to fulfill reporting obligations. Along with the development of intelligent connected vehicles, accident reporting and analysis are gradually becoming important content of regulation.
Xiao Lingyun stated at the conference that the vehicle recall regulation is being revised. For smart cars, the institutional revision needs to adapt to new technical architectures and industrial collaboration methods, ensuring that defects can be discovered in a timely manner, responsibilities can be accurately determined, and relevant enterprises can fulfill their rectification and recall obligations according to the law.
Forward-Shifting Regulation and Targeting Overseas Markets
As the recall system is being revised, regulatory departments are also attempting to extend the identification of quality risks forward. Xiao Lingyun summarized the current safety regulation as "three lines": market access is the baseline, sandbox regulation is the elevated line, and recalls are the bottom line; the three lines are implemented together.
Market access mainly controls the gateway for products entering the market, while recalls are responsible for taking disposal measures after defects are discovered. Sandbox regulation is a newly added regulatory tool in the automotive industry in recent years.
In standard scenario validations like those in laboratories, the system's ability to identify and process irregular objects and non-standard scenarios still requires continuous validation. The role of sandbox regulation is, before and after vehicles enter the market, to help enterprises discover risks easily missed in routine validation through tests close to real-world usage scenarios.
Xiao Lingyun introduced that currently, over 2 million mass-produced vehicles have been improved through sandbox regulation, avoiding large-scale recalls in the aftermarket. New smart models from mainstream automakers are also successively participating in sandbox testing.
Complementing the sandbox is the accident reporting system. The State Administration for Market Regulation (SAMR) and the Ministry of Industry and Information Technology (MIIT) jointly introduced relevant requirements, mandating that enterprises must report on time after intelligent driving-related accidents occur. Regulatory departments aggregate accident data, establish a database of dangerous scenarios, and then use AI large models to automatically generate new testing scenarios from historical accidents, pushing them back to automakers for validation.
This mechanism attempts to transform the problems exposed in accidents into content for subsequent testing. Currently, seven or eight enterprises are using these automatically generated scenarios for sandbox testing. Once this data closed-loop is fully operational, regulatory departments can proactively determine which designs pose risks.
In addition to accident monitoring, regulatory departments are also improving standards in software, power batteries, and overall vehicle validation.
The first domestic standard for automotive software quality and defect management has been released, filling the gap in the field of software quality management. In terms of power batteries, the Recall Center and enterprises such as CATL have successively released multiple national standards for quality management last year.
The validation phase is also in motion. In August, four departments launched a one-year special action to improve the production consistency and quality of road motor vehicle products, with reliability, durability, and new technology testing validation listed as key focuses. Three finalization test procedures for electric vehicles are also soliciting opinions, with the total reliability driving mileage uniformly increased from the original approximately 15,000 kilometers to no less than 30,000 kilometers, aligning with ICE vehicle standards.
Gong Jinfeng, General Manager of the China Automotive Technology and Research Center (CATARC), proposed a suggestion in the same direction at the opening ceremony: improve the full-process consistency control system from core component procurement, production process control, to product factory inspection, and promote the transformation of industry competition from price comparison to quality upgrading.
Enterprises are also responding. Seres proposed the "Five Highs" new standards: high safety, high reliability, high performance, high quality, and high value. GAC Group publicly stated after the launch of the special action: "Resolutely refuse to build quick-built cars, only put in hard work."
From development validation to accident monitoring, and then to defect investigation and recall disposal, the quality management of regulators and enterprises is covering a longer product lifecycle. For automakers, product launch is just a node; subsequent usage performance, problem identification, and rectification capabilities are equally related to product quality and user safety.
The regulatory vision is also extending overseas.
From January to July 2026, China's automobile exports reached 6.41 million units, a year-on-year increase of 54%. As the export scale expands, the market access regulations, certification requirements, and recall rules in overseas markets have also become practical issues that China's automotive industry needs to face.
The Recall Center and CATARC have cooperated to build the "China Automotive Overseas Quality Standard and Regulation Technical Service Platform", covering the main destination countries for Chinese car exports. Xiao Lingyun said that building this platform aims to help Chinese cars encounter fewer compliance issues, as well as fewer quality guarantee and recall problems overseas.
For China's automotive industry, competition in overseas markets has extended to multiple links such as product quality, regulatory adaptation, and after-sales disposal. Enterprises need to consider the requirements of different markets during the product development phase, continuously track quality performance after vehicle delivery, and possess the ability to handle defects and fulfill recall obligations.
Xiao Lingyun stated, "Safety truly defines our cars. Just as surging waves sift the sand, what remains will undoubtedly be brands and models that excel in both quality and safety."