China Releases Seven National Standards for Interconnection of AI Agents to Break Down Industry-Wide Information Silos of AI Models

China National Radio, Beijing, June 26 (Reporter Shao Lanjie) China’s AI agent industry is experiencing rapid growth with expanding application scenarios. As cross-platform, cross-scenario and cross-domain interconnections become increasingly prevalent, the problem of "information silos" among AI agents has become prominent. On June 26, reporters learned from the State Administration for Market Regulation (SAMR) that a series of national standards titled Artificial Intelligence – Interconnection of AI Agents have been issued, which are expected to resolve prominent pain points including inconsistent communication interfaces and the lack of an identity management system for interconnected AI agents.

 

A total of seven national standards under the Artificial Intelligence – Interconnection of AI Agents suite have been approved and released. They comprehensively cover core links including general architecture, identity codes, identity management, agent description, agent discovery, agent interaction and agent tool calling. Together, they systematically build a closed-loop standard system covering the full workflow: identity identification, capability description, supply-demand matching, collaborative interaction and tool invocation, filling major gaps in standards for this sector.

 

Zhu Meina, Deputy Director-General of the Standards Technical Department of SAMR, stated that relevant standards for identity codes will be converted into mandatory national standards in due course. Meanwhile, the formulation of standards for niche segments such as AI agent auditing and AI agent transactions will be accelerated, establishing a compliant, efficient, secure and credible rule framework for the sharing of agent capabilities and circulation of agent value.

 

Fan Kefeng, Vice President of China Electronics Standardization Institute, said the suite of interconnection standards sets fundamental principles and capability access thresholds for AI R&D and industrial application, delivering three core values: First, breaking silos to build networks. The standards transform isolated closed agent systems into interconnected nodes, fostering a networked intelligent ecosystem. Second, shifting from proprietary to open architecture. Unified description and discovery mechanisms greatly lower technical barriers for small and medium-sized enterprises, enabling them to integrate into mainstream ecosystems and stimulate innovation across long-tail market participants. Third, turning disorder into standardized order. At a critical stage where autonomous agent capabilities are advancing rapidly and deeply penetrating physical scenarios, the standards proactively define boundaries for secure and compliant interaction, balancing industrial innovation with safety governance and laying a foundational framework for the safe, trustworthy, ethical and controllable development of artificial intelligence.

 

Regarding the implementation of the new standards, Tang Chao, Deputy Director of Zhongguancun Science City Administrative Committee and Vice District Chief of Haidian District, noted that many local enterprises including Volcano Engine, Xiaomi, Kuaishou and Lenovo deeply participated in drafting the seven national standards, integrating Haidian’s frontline industrial practices into the national standard system to ensure practicality and implementability. Moving forward, targeted standard compliance pilot projects will be launched around new AI business forms and models to accelerate ecosystem expansion.