On August 25, 2026, the China National Institute of Standardization (CNIS) issued an analysis for GB/T 47863—2026 Social impact of generative artificial intelligence technology application—Service providers compliance management guideline (hereinafter referred to as “the Standard”) a national standard that was issued in July of 2026 and will take effect on November 1 of the same year. It was proposed and administered by the SAC/SWG35 (Social Application and Evaluation of Intelligent Technology), and the main drafting organizations include CNIS, Tsinghua University etc.

Generative AI has been widely used in economic, judicial, government affairs, education, medical care, scientific research, and other fields both globally and in China. It not only improves decision-making efficiency and optimizes organizational models, but also profoundly affects production, life, and social governance, and causes shocks to traditional trust mechanisms, knowledge production mechanisms, and rights generation mechanisms. For the standardization community and policy makers in China, how to identify in advance, effectively assess, and guide technology applications to produce benign and positive social impacts has become an urgent issue. The Standard intends to provide guidance for service providers to establish a comprehensive and effective social impact compliance management system, and applies to their related compliance management activities.

From the main text of the standard, it does not set up a separate independent system. Instead, based on GB/T 35770—2022 Compliance management systems—Requirements with guidance for use (IDT, ISO 37301:2021), and combined with the characteristics of generative AI, it is usually embedded in an existing compliance management system or artificial intelligence management system as a special module or extended application. The Standard requires service providers to understand their own environment, the needs and expectations of interested parties, determine the scope of the system, and clarify compliance management processes. Governance bodies and top management should commit, formulate policies, allocate resources, and establish accountability mechanisms; compliance management bodies, business departments, and all personnel have their respective responsibilities. In the planning stage, they should determine compliance objectives, identify compliance obligations, and carry out risk assessment and response. The support level involves resources, competence, communication and collaboration, and documented information. The operation level includes management plans, compliance review, compliance consultation, and compliance whistleblowing and investigation. Evaluation and improvement cover situation monitoring, performance evaluation, effectiveness evaluation, compliance reporting, and improvement.

The appendices of the standard generally summarize common compliance obligations and non-conforming or non-compliant phenomena. The former includes three categories: technology and qualifications, service provider behavior, and generated content; the latter are listed at three levels: individuals and groups, organizations and industries, and economic and social systems, such as privacy leakage, minors' addiction, dissemination of false information, algorithmic bias, and energy consumption.

Its relationship with other standards is also relatively clear. This document is aligned with GB/T 46800—2025 Social impact of generative artificial intelligence technology application—Evaluation guidelines, and the two together provide standardized support for social impact governance; it normatively references GB/T 35770 and GB/T 46800, and its bibliography also involves artificial intelligence management systems, labeling of AI-generated synthetic content, basic security requirements for generative artificial intelligence services, and so on.

Its importance lies in that the standard incorporates social impact into compliance management, helps service providers carry out activities in accordance with laws and regulations, addresses data security, privacy risks, and ethical challenges, promotes the healthy development and standardized application of generative AI, safeguards national security and public interests, protects the lawful rights and interests of citizens, legal persons, and other organizations, and enhances safety and reliability.