China launches identity registry for AI agents to ensure trust

A new framework for AI identity

China recently introduced a dedicated identity registration and resolution platform for AI agents during the World Internet Conference Wuzhen Summit. This initiative was presented during a forum focused on how these autonomous systems can interact and maintain trust across different networks. The platform is designed to handle identity coding and cross-domain resolution, which allows different systems to recognize and verify the specific AI agents they are communicating with.

The project is being built by national-level parties to create a foundation for trusted interoperability. This means that different AI agents will be able to work together seamlessly while following a unified set of rules. Operating platforms will be layered on top of this infrastructure to host the actual access points and the publishing of various agent capabilities.

A pilot program is already underway to test how these operating platforms and service providers can connect. This early phase has already successfully linked several providers to test the feasibility of the registry. The goal is to move beyond simple content moderation to a system where the actors themselves are verifiable and their behaviors are traceable.

Industry leaders sign on to rules

Twenty major organizations joined the initiative by releasing a self-discipline convention for AI agent development. This group includes some of the largest technology players in the region, such as Huawei, Alibaba, Tencent, Xiaomi, and Baidu. The convention was drafted by the China Academy of Information and Communications Technology alongside several other tech giants including Volcano Engine, Ant Group, and Douyin.

The agreement outlines eight specific commitments that the participating companies must follow. These commitments cover a wide range of areas, including trust systems, interconnection infrastructure, and security protection. They also address governance mechanisms, user rights, and the importance of open sharing of technology within the industry.

Signatories have pledged to develop specific technologies for rule-embedding and compliance testing. They also committed to working together to prevent the use of AI agents for fraud or algorithmic exploitation. By signing this convention, these companies are agreeing to accept oversight from both industry bodies and the general public to ensure their agents behave responsibly.

How the system works in practice

The infrastructure for these agents relies on shared standards for registration and capability discovery. This allows an AI agent to

Expanding the trust infrastructure

Chen Liang, who serves as the vice president of Ant Group and the chief technology officer for its security and Alipay business group, emphasized the necessity of this new framework. He argued that the upcoming era of AI agents requires a trust infrastructure that embeds identity, authorization, and responsibility into every single connection made by these systems. This approach ensures that every interaction between an agent and a service is governed by clear permissions and accountability.

To support these goals, Ant Group is leveraging its APASS unified identity account as part of its strategy. The company is also developing the ASL protocol in collaboration with more than 20 different partners. These tools are intended to provide the underlying technical standards needed to manage how agents move through different digital environments while maintaining a secure and verifiable presence.

Shifting the focus of online governance

The forum where these developments were announced was hosted by the GSMA and the Cyberspace Administration of China. During the event, Wang Jingtao, the deputy director of the Cyberspace Administration of China, called for the establishment of basic rules regarding agent identity and authorization management. He noted that the goal is to extend current online governance beyond just content and model security to include the verification of actors and the traceability of their behaviors.

This shift in policy suggests a move toward monitoring the actual actions and responsibilities of AI entities rather than just the data they produce. By focusing on compliant operation and verifiable actors, the government aims to create a more controlled environment for autonomous systems. This strategy seeks to ensure that as agents become more capable, their actions remain traceable and subject to oversight by both industry bodies and the public.