China’s Ministry of Industry and Information Technology (MIIT ) has officially released a mandatory national safety standard for advanced driver-assistance systems (ADAS), aiming to regulate the rapidly expanding integration of artificial intelligence in consumer vehicles. The new standard, designated GB 47955-2026, will take effect on January 1, 2027, and establishes strict testing and operational requirements for automakers deploying these technologies on public roads. The regulation is designed to curb the overhyping of autonomous capabilities and mitigate the risks associated with drivers over-relying on systems that still require human supervision.
According to Caixin Global, the standard covers three primary categories of assisted driving: basic single-lane systems, basic multi-lane systems, and more advanced navigation-assisted systems, commonly referred to as Navigate on Autopilot (NOA). The mandate requires automakers to conduct rigorous closed-course testing, implement mandatory driver education protocols, and use continuous driver attention-monitoring systems. Furthermore, vehicles must be equipped with key data recording capabilities to facilitate accident investigation and liability determination.
Addressing Safety Risks and Overhyped AI Claims
The implementation of GB 47955-2026 comes in response to several high-profile accidents involving ADAS-equipped vehicles in China between 2025 and 2026. In many of these incidents, drivers mistakenly believed their vehicles were fully autonomous, leading to fatal lapses in attention. By mandating continuous driver monitoring and strict educational requirements, the MIIT is attempting to force a clear distinction between Level 2 assisted driving, which currently represents the most advanced technology available in Chinese smart vehicles, and true Level 4 or Level 5 autonomy.
Industry experts, such as Liu Hui, note that while the new standard will inevitably raise research, development, and testing costs for automakers, it will also compel the industry to abandon inadequate or unsafe technical solutions. This regulatory floor is expected to trigger market consolidation, favoring well-capitalized companies with robust AI engineering capabilities over smaller players that may have rushed immature systems to market to capitalize on consumer demand for “smart” features. The requirement for continuous driver attention monitoring, in particular, will require significant hardware and software investment from automakers that have not yet deployed such systems.
The standard also has important implications for the competitive dynamics between domestic and foreign automakers in China. Chinese EV manufacturers like BYD, Li Auto, and NIO have been aggressive in marketing their AI-powered driving features, often using language that implies a higher degree of autonomy than the technology currently supports. GB 47955-2026 will require all of these companies to align their marketing claims with the technical reality of Level 2 systems, potentially tempering some of the consumer enthusiasm that has driven strong sales.
Global Context and the Road to Full Autonomy
China’s move to standardize ADAS safety aligns with broader international efforts to govern automotive AI. A United Nations global regulation on autonomous driving, co-drafted by China alongside the EU, UK, US, Canada, and Japan, was recently approved, indicating a consensus on the need for harmonized safety frameworks. However, GB 47955-2026 specifically targets assisted driving, acknowledging that fully autonomous vehicles are not yet ready for unregulated consumer deployment.
Looking ahead, the MIIT is already preparing the next phase of regulatory oversight. A draft national standard specifically addressing fully autonomous systems (Level 4 and above) is expected to be released by July 1, 2027. Until then, the current standard serves as a critical bridge, ensuring that the rapid deployment of AI in the automotive sector does not outpace the fundamental requirements of public safety. By establishing clear technical and operational boundaries, Beijing is attempting to foster sustainable innovation while protecting consumers from the bleeding edge of AI development. The January 2027 implementation date gives automakers approximately six months to bring their systems into compliance, a timeline that industry observers describe as achievable but demanding.
For international automakers operating in China, GB 47955-2026 presents both a compliance burden and a competitive opportunity. Foreign brands that have already invested heavily in robust ADAS safety architectures in their home markets may find it relatively straightforward to meet the new Chinese requirements, while domestic competitors that have prioritized feature-richness over safety rigor may face more significant engineering challenges. The standard’s data recording requirements, in particular, are likely to generate a substantial new market for automotive-grade data storage and telematics solutions, creating opportunities for component suppliers across the supply chain.
The release of GB 47955-2026 also signals a broader shift in Beijing’s view of the relationship between AI innovation and public safety. For much of the past five years, Chinese regulators adopted a permissive stance toward automotive AI, prioritizing the growth of the domestic EV and smart vehicle industry over prescriptive safety requirements. The new mandatory standard marks a transition to a more mature regulatory posture, one that acknowledges the technology has reached sufficient scale and market penetration to warrant formal governance. This evolution mirrors the trajectory of AI regulation in other domains, such as generative AI and facial recognition, where China has moved from initial permissiveness to structured oversight as technologies become embedded in daily life.
(Related: Baidu and Lyft to Launch Robotaxi Testing in London, Marking Apollo Go’s First European Deployment)