China proposes a global rulebook for agent identity and interoperability
The CAC's ten-pillar initiative calls for agent identity systems and open interop standards at the interface, protocol, semantic, and process layers. Translations landed this weekend.
- The Cyberspace Administration of China released a Global Cooperation Initiative on agent mutual trust, interconnection, and interoperability at WAIC, its first systematic position on AI agent interop.
- The ten-pillar blueprint calls for agent identity identification, mutual recognition systems, and open international standards across interface, protocol, semantic, and process layers.
- Full English translations and international analysis circulated over the weekend, and the initiative courts Global South signatories with pledged support for developing economies.
The protocol wars just acquired a state actor with a ten-point plan. The Cyberspace Administration of China released its Global Cooperation Initiative on Agent Mutual Trust, Interconnection, and Interoperability at the World Artificial Intelligence Conference in Shanghai, and the document's full text and English translations began circulating internationally this weekend, per Geopolitechs, which published a detailed analysis.
It is the first systematic articulation of China's position on AI agent interoperability, and it is comprehensive: ten pillars spanning technology, standards, infrastructure, applications, industry, security, data, resources, culture, and literacy. The framing is a triad, with mutual trust as the foundation, interconnectivity as the channel, and interoperability as the rules of engagement.
Identity first, standards second
Two provisions matter most for this beat. The initiative calls for research into "agent identity identification" and mutual recognition systems so autonomous agents can verify each other before cooperating, the same problem the ITU began formalizing this month and Entrust is selling into enterprises. And it demands "open, non-discriminatory, and transparent international interoperability standards" across the interface, protocol, semantic, and process layers, a four-layer cut that maps loosely onto where MCP and A2A already operate.
A counterpoint to industry-led standards
The subtext is who writes the rules. Western agent standards have been industry led, with A2A under the Linux Foundation and MCP governed by its maintainers. China's initiative is state led, arrives alongside the new WAICO alliance headquartered in Shanghai, and is explicitly pitched at Global South governments with support pledged for developing economies. Standards bodies move slowly and agent deployments do not, so the near-term effect is not a treaty but a second gravitational center. Interop was already a technical race. As of this weekend it is a diplomatic one too.