On 5 June 2024 the Ministry of Industry and Information Technology, the Cyberspace Administration of China, the National Development and Reform Commission and the Standardisation Administration jointly issued the Guide to the Construction of a Comprehensive Standardisation System for the National Artificial Intelligence Industry (2024 Edition). The guide divides the system into seven parts: foundational common standards, foundational support, key technology, intelligent products and services, enabling new industrialisation, industry applications, and safety/governance. Its value is not that another document exists, but that a company can use it to locate the standards it should follow.
How the seven parts map to company activity
Foundational common standards cover terminology, reference architecture and test methods. Foundational support covers compute, data and toolchains. Key technology points to algorithms and model capability. Intelligent products and services correspond to devices and applications. Enabling new industrialisation and industry applications address deployment scenarios. Safety/governance runs across the whole system. A company building foundation models and a company building sector applications do not share the same priorities.
Standards, filings and assessments
Most standards are recommended rather than mandatory, but they are often cited as reference points in filing, safety assessment, government procurement and sector access. Understanding them helps predict what a reviewer will focus on and makes technical documents more targeted. Turning standard requirements into an internal checklist is more practical than checking clauses at short notice. For companies submitting algorithm filings or safety assessments, terminology and test methods in the standards often shape how the material should be written.
How to adopt standards
First identify the relevant family of standards, then distinguish mandatory from recommended ones, and finally map key clauses to product documents, test reports and management rules. For small and medium AI companies, addressing standards on safety, data and labelling first usually produces the highest return. Adoption is not a one-off certification but a process maintained alongside product versions; when a standard is revised, referenced clauses should be rechecked.
Mandatory versus recommended standards
The seven-part system describes a standardisation plan, not a set of uniformly binding rules. Companies need to separate mandatory national standards, recommended national standards and industry standards, and then determine which enter regulatory requirements by reference. Treating every recommended standard as an obligation raises cost; treating a mandatory standard as guidance creates substantive risk. The test is the nature of the standard and the provisions of the regulatory document that applies. Companies serving several sectors should also map references sector by sector rather than applying one set everywhere.
The value of participation
Participating in drafting can turn a company’s technical practice into industry language and provide early warning of direction changes. Participation is not influence, but it at least reduces the cost of passive adaptation. For outbound companies, domestic standard practice can also be a basis for dialogue with international standards bodies. Whether to invest depends on the company’s stage and product lifecycle; it is not a universal answer.
For standards review and filing alignment, see our algorithm filing service and data compliance service; definitions are collected in the glossary.
This article is based on the Guide to the Construction of a Comprehensive Standardisation System for the National Artificial Intelligence Industry (2024 Edition) jointly issued by the Ministry of Industry and Information Technology, the Cyberspace Administration of China, the National Development and Reform Commission and the Standardisation Administration on 5 June 2024, and on public information. It is for reference only and does not constitute legal advice.