On November 6, 2026, China's Ministry of Industry and Information Technology published a draft regulation that would give every robot manufactured in the country a unique identity code. The code has four parts: a manufacturer code, a product-description code, a product-sequence code, and an "other" segment. It must be affixed before the robot leaves the factory, in at least three forms — digitally stored, human-readable, and machine-readable. It must be "clear, visible, durable, not easily damaged, not easily replaced." It must remain unchanged for the life of the machine: not tamperable, not erasable.
The public comment period runs through December 5. The underlying national standard, GB/T 48541-2026, was issued on October 30 and takes effect on February 1, 2027. The regulation applies to industrial robots, service robots, and special-purpose robots — every category, not just the humanoid machines that received a similar code in an earlier pilot. By May 2026, that humanoid pilot had already assigned 29-character codes to more than 28,000 robots across 100-plus companies, using a four-part structure of country code, enterprise code, product code, and serial number.
Three weeks earlier, on November 3, the National Development and Reform Commission held a press conference on the 15th Five-Year Plan. Wang Ruomeng, a deputy director in the commission's innovation and high-tech department, described the government's approach to AI governance in four points. The first was to "accelerate foundational and comprehensive AI legislation" and to "establish safety standards and ethical guidelines covering the full lifecycle of artificial intelligence" — spanning data, computing power, algorithms, and key application scenarios.
Two announcements, three weeks apart, describing the same architecture from two directions. The NDRC is building the legal frame. MIIT is building the enforcement mechanism that makes the frame operational at the level of a single machine.
That combination is not how the American AI governance debate has unfolded. The United States has focused on voluntary frameworks, pre-deployment testing agreements with frontier labs, and a classified benchmarking process for cyber capability. China is building something different: a standards infrastructure that reaches from the data center to the factory floor, and from the training run to the resale listing of a used robot.
What "Full Lifecycle" Means in Practice
The phrase appears in almost every Chinese policy document about AI. The robot code regulation shows what it means when translated into an enforceable rule.
A robot is assigned a code before it leaves the factory. The manufacturer must apply for the code through a service platform, and the application volume is capped at 300 percent of the previous year's actual output. The cap is designed to prevent manufacturers from stockpiling codes or inflating production figures. Every assigned code must be filed with the platform within 20 working days of being affixed. The filing includes structural model, power source, and other details that are held by a third-party technical support institution and made queryable by users.
The code follows the machine through its entire existence. When the robot is sold, the buyer can query its manufacturing history. When it is repaired, the repair record is tied to the code. When it is resold on the secondhand market, the buyer can see its usage history and maintenance record rather than relying on the seller's description. When it is retired and recycled, the code closes the loop.
The regulation also reaches backward. Robots that were manufactured and sold before the rule takes effect must have their manufacturer-defined serial numbers filed with the platform. China's industrial robot installed base is among the world's largest, according to industry data cited by Business Times China. The humanoid pilot platform alone has registered more than 28,000 machines. The all-category rule would bring the broader installed base into the same traceability system.
That is what "full lifecycle" means when it is operationalized. It is a data structure that makes every physical AI system identifiable, traceable, and accountable from the moment of manufacture to the moment of disposal, rather than a principle about safety alone.
The Standards Stack Behind the Code
The robot code regulation is one node in a much larger standards infrastructure that China has built over the past two years.
The coverage is not limited to model performance. It spans the full chain: basic software and hardware, key technologies, industry applications, and safety governance. The standards include technical requirements for retrieval-augmented generation, specifications for embodied intelligence systems, safety requirements for government large-model systems, and lifecycle security norms for AI agents.
The AI agent standards illustrate the same logic as the robot code. In July 2026, a set of seven national guiding standards — GB/Z 185-2026 — was published, forming a "full lifecycle closed loop" across architecture, identity, trusted management, capability description, intelligent discovery, multi-modal interaction, and tool invocation. An earlier set of implementation opinions on AI agent governance, issued by three ministries in May 2026, had called for "full-cycle safety norms covering agent development, deployment, application, and maintenance," with security management at the points of model access, API invocation, and tool use.
The robot code is the physical-world counterpart of the agent standards. An AI agent operating in software has an identity that can be registered, authenticated, and traced. A robot operating in the physical world has a code that can be affixed, queried, and followed from factory to scrapyard. The two systems share the same architecture: identity, traceability, lifecycle accountability.
The scale of the standards effort is visible in the publication cadence. On September 10, 2026, the Standardization Administration released 298 national standards in a single batch, 47 of which were AI-related, according to 36Kr's reporting. In August 2026, MIIT Vice Minister Xin Guobin said China had formulated nearly 200 key AI standards, according to Xinhua and Guangming Daily.
The Enforcement Logic
The distinction between a standard and an enforcement mechanism is the distinction between a principle and a code that runs.
China's AI standards are, in form, largely recommended rather than mandatory. GB/T, the prefix on the robot code standard, denotes a recommended national standard. Companies can voluntarily adopt it. But when a recommended standard is cited in an administrative regulation, it acquires de facto binding force within the scope of that regulation. The robot code interim measures cite GB/T 48541-2026 explicitly. The standard is nominally voluntary, but the regulation that cites it is not.
That mechanism turns the standards stack into an enforcement system. The NDRC builds the legislative frame. MIIT issues the administrative measures. The standards provide the technical specifications. The platform provides the registration and query infrastructure. The manufacturer bears the cost of compliance.
The cost is not trivial for smaller firms. A production line that has never applied a machine-readable code to every unit must invest in marking equipment and system integration. The 300 percent cap on code applications constrains production planning. The 20-day filing deadline requires a compliance workflow. And the regulation does not yet clarify whether a repaired robot needs a new code or an amended filing — a detail that matters for machines with high failure rates.
But the compliance burden is the point. A system that imposes no cost on the manufacturer is a system that cannot enforce anything. The robot code imposes a cost at the point of production, and in exchange it creates a record that follows the machine for its entire life.
The Contrast With the American Approach
The United States has not built an equivalent system. It has built a different kind of system, aimed at a different problem.
The American approach to AI safety has centered on frontier models. The Commerce Department's Center for AI Standards and Innovation has signed pre-deployment evaluation agreements with the major labs. The June 2026 executive order established a classified benchmarking process for cyber capability. The Treasury Department's outbound investment rules restrict capital flows to Chinese AI and quantum companies. The State Department has sent letters to allied countries warning them against adopting Chinese AI infrastructure.
These are instruments of supply control and frontier review. They aim to restrict who can build the most advanced models, who can invest in them, and which countries can deploy them. They do not aim to give every robot in the country a unique, tamper-proof identity code.
The American system also lacks a unified federal AI statute. Governance is distributed across executive orders, agency rules, state legislation, and litigation. There is no equivalent of the NDRC's "foundational and comprehensive AI legislation," and no equivalent of MIIT's all-category robot registry.
The two systems are not competing to solve the same problem. The United States is trying to control the frontier. China is trying to instrument the base.
What the Robot Code Reveals
The robot code regulation is a small document. It governs a serial number. But it reveals the shape of China's AI governance strategy more clearly than any white paper.
The strategy is to build the identity layer of physical AI before the market scales. A robot without a code can be sold, repaired, resold, and scrapped without any record that follows it. A robot with a code is a node in a data structure. The manufacturer is known. The product specifications are known. The repair history is known. The current owner can be queried. The resale value depends on the record.
That data structure is the precondition for everything else the policy intends to do: safety governance, liability assignment, secondhand market regulation, recycling, and the kind of "full lifecycle" oversight that the NDRC described.
The United States has not built this layer. It has built a frontier-review layer and a capital-control layer. Those layers matter for national security. They do not tell you which robot is in which factory, who made it, when it was repaired, or where it went after it was sold.
China's bet is that the identity layer is the more durable form of governance. A frontier model can be restricted at the API. A robot with a code is restricted at the point of existence. The first can be routed around. The second is inscribed on the machine.
The comment period ends December 5. The standard takes effect February 1. The robots are already being coded.
Sources: National Development and Reform Commission press conference, November 3, 2026, via Beijing Daily (November 3, 2026), Tencent News (November 6, 2026), Eastmoney (November 4, 2026), and Bloomberg via Gateweb3 (November 4, 2026); MIIT draft interim measures on robot product identification codes, November 6, 2026, via MIIT Equipment Industry Development Center (November 6, 2026), Tencent News (November 5, 2026), Pandaily (November 7, 2026), and IT Home (November 7, 2026); GB/T 48541-2026 standard issuance and February 1, 2027 effective date via Business Times China (November 6, 2026); humanoid robot lifecycle platform covering 100-plus companies and 28,000-plus robots via IT Home (November 7, 2026) and Pandaily (November 7, 2026); 300 percent application cap and 20-day filing requirement via Business Times China (November 6, 2026); MIIT Vice Minister Xin Guobin on nearly 200 AI standards, August 26, 2026, via Xinhua and Guangming Daily; 298 national standards including 47 AI-related standards via 36Kr (September 11, 2026); AI agent full-cycle safety norms via Cyberspace Administration, NDRC, and MIIT implementation opinions (May 8, 2026); seven-part national standard series GB/Z 185-2026 on AI agents via People's Daily (July 11, 2026); U.S. CAISI pre-deployment evaluation agreements and June 2, 2026 executive order via Commerce Department and DLA Piper analysis (2026).
Disclaimer
The information provided in this article is for general informational and educational purposes only. It does not constitute legal, financial, or professional advice. The author and publisher are not responsible for any actions taken based on the content of this article. Readers should consult qualified professionals for advice specific to their situation. All trademarks and references to third-party products, services, or organizations are the property of their respective owners. The performance data and benchmarks discussed are based on specific research studies and may not generalize to all use cases or environments. As of the publication date, the AI landscape continues to evolve rapidly, and readers should verify current information independently.
Limitations
The robot identity code rule is a draft released for public comment and may change before it takes effect.
Deadlines described are those stated in the draft documents and have not been confirmed by final implementing rules.
Chinese regulatory text was read in the original and rendered in English here; official translations may differ in terminology.
The scope of covered robot categories and the enforcement mechanism for the code are subject to further guidance.
Reporting standards differ between Chinese and Western outlets, and figures cited here reflect the sourcing available at the time of writing.
Market size and shipment estimates vary by methodology; where estimates conflict, this article cites the figure used by the original source.
The sectors discussed are changing quickly, and some figures may be revised after publication.
Sources
- National Development and Reform Commission press conference, November 3, 2026, via Beijing Daily (November 3, 2026), Tencent News (November 6, 2026), Eastmoney (November 4, 2026), and Bloomberg via Gateweb3 (November 4, 2026)
- MIIT draft interim measures on robot product identification codes, November 6, 2026, via MIIT Equipment Industry Development Center (November 6, 2026), Tencent News (November 5, 2026), Pandaily (November 7, 2026), and IT Home (November 7, 2026)
- GB/T 48541-2026 standard issuance and February 1, 2027 effective date via Business Times China (November 6, 2026)
- humanoid robot lifecycle platform covering 100-plus companies and 28,000-plus robots via IT Home (November 7, 2026) and Pandaily (November 7, 2026)
- 300 percent application cap and 20-day filing requirement via Business Times China (November 6, 2026)
- MIIT Vice Minister Xin Guobin on nearly 200 AI standards, August 26, 2026, via Xinhua and Guangming Daily
- 298 national standards including 47 AI-related standards via 36Kr (September 11, 2026)
- AI agent full-cycle safety norms via Cyberspace Administration, NDRC, and MIIT implementation opinions (May 8, 2026)
- seven-part national standard series GB/Z 185-2026 on AI agents via People's Daily (July 11, 2026)
- U.S. CAISI pre-deployment evaluation agreements and June 2, 2026 executive order via Commerce Department and DLA Piper analysis (2026).
The information provided in this article is for general informational and educational purposes only. It does not constitute legal, financial, or professional advice. The author and publisher are not responsible for any actions taken based on the content of this article. Readers should consult qualified professionals for advice specific to their situation. All trademarks and references to third-party products, services, or organizations are the property of their respective owners. The performance data and benchmarks discussed are based on specific research studies and may not generalize to all use cases or environments. As of the publication date, the AI landscape continues to evolve rapidly, and readers should verify current information independently.
Limitations: The robot identity code rule is a draft released for public comment and may change before it takes effect.; Deadlines described are those stated in the draft documents and have not been confirmed by final implementing rules.; Chinese regulatory text was read in the original and rendered in English here; official translations may differ in terminology.; The scope of covered robot categories and the enforcement mechanism for the code are subject to further guidance.; Reporting standards differ between Chinese and Western outlets, and figures cited here reflect the sourcing available at the time of writing.; Market size and shipment estimates vary by methodology; where estimates conflict, this article cites the figure used by the original source.; The sectors discussed are changing quickly, and some figures may be revised after publication.