繁體中文RSS

Technology & Public Safety

China Issues Its First Mandatory Safety Standard for Self-Driving Cars

A new mandatory national standard, GB 44721-2026, creates a unified safety baseline for L3 and L4 autonomous driving systems in China, set to take effect in July 2027.

As self-driving technology moves from test tracks toward commercial roads, the question of how to keep it safe has become central. China has now answered part of that question with a landmark rule: GB 44721-2026, the country's first mandatory national standard for autonomous driving system safety. Approved by the State Administration for Market Regulation and the Standardization Administration, it was released in early August 2026 and is scheduled to take effect on July 1, 2027.

The timing reflects a fast-moving industry. In 2025, two L3-grade (conditional automation) models were granted conditional market-access permits in China, signaling that partially self-driving cars were ready to leave the laboratory. The new standard aims to keep pace with that momentum by giving regulators and manufacturers a common, enforceable reference point for what counts as a safe autonomous driving system.

The standard applies to M and N class vehicles — passenger cars, buses, and trucks — that carry L3 or L4 automated driving systems. Notably, it does not cover automatic parking. Rather than prescribing a single technology, it builds a complete safety framework tailored to China's industrial and regulatory environment, balancing technical feasibility with real-world practicality.

At the core of the rule is a whole-lifecycle safety approach. Manufacturers must establish mechanisms that cover the autonomous system from design through operation, addressing potential risks at the source. During the research and development phase, they are required to organize simulation-based tests, proving-ground trials, and on-road testing. The standard also demands that the system handle dynamic driving tasks reliably and that the activation and exit processes be safe, with clear rules for signaling the system's readiness, activation, and deactivation.

One of the most important requirements concerns the human driver. For L3 systems, where the driver is expected to remain available to take over, the vehicle must include a driver-takeover monitoring capability. This ensures the system can confirm that the person behind the wheel is ready to resume control before handing it back — a critical safeguard during the transition between machine and human driving.

The standard also fleshes out multi-dimensional testing and inspection methods so that compliance can actually be verified. For the industry, this provides a degree of certainty that has been missing: a clear, mandatory baseline against which autonomous vehicles can be assessed. For consumers, it means that when self-driving features arrive on the market in earnest, there will be a defined safety floor beneath them, rather than a patchwork of voluntary guidelines.