China has unveiled a platform intended to give AI agents a way to identify and connect with services across platforms, while 20 organisations signed a pact on agent development. The practical test is already under way: a pilot has linked several platforms, including Alipay services, through trusted connections.
Watch Desk analysis
What happened
The platform was unveiled at the World Internet Conference’s Wuzhen Summit on 4 November. PanDaily, citing reports from IT Home and China Daily’s Chinese-language site, says it is being built to handle agent identity coding and cross-platform resolution, with operating platforms providing access to agents and their capabilities.
The pilot has connected several platforms and service providers. Ant Group is among its early participants; the reports say an Alipay government-services agent and Sesame Credit have completed trusted links with agents on other platforms. The platform is still being built out, so this is an early test rather than evidence of a finished, widely available system.
Alongside the unveiling, 20 organisations including Huawei, Alibaba, Tencent, Xiaomi and Baidu released a self-discipline convention. Its eight commitments cover agent identity and interoperability, security, governance, user rights and industry adoption. Signatories also pledge to develop compliance-testing technology and resist using agents for fraud or algorithmic exploitation.
Why it matters
Agents that can move between services need more than a clever model. They need ways to identify themselves, find other services and establish what they are allowed to do. The platform’s pilot, including links to named Alipay services, gives the announcement a more concrete edge than a list of principles alone.
But shared infrastructure and a voluntary pact are not the same as proven safeguards. The important questions are how identity and authorisation work in practice, who checks compliance, and what happens when an agent misbehaves across platform boundaries.
Our read
This is a meaningful attempt to build the plumbing for agents to work across services, with an early pilot to show where the pipes might lead. The convention is a useful signal of industry coordination, not proof that users already have reliable protections. The detail to watch is whether the technical rules make permissions and responsibility verifiable, rather than merely agreeable on a conference stage.
What to watch
- Whether the platform expands beyond the current pilot and publishes its technical standards.
- How agent identity, authorisation and responsibility are checked across different platforms.
- Whether the convention’s security and user-rights promises become testable requirements.
Discussion spark: Should cross-platform AI agents need a shared, verifiable identity and permission system before they can act, or would that give the platforms too much control over who gets to build them?
Sources and evidence
- China Unveils an AI Agent Identity Registry at Wuzhen as 20 Organizations Sign Self-Discipline Pact (6 November 2026, 01:50 UTC)
- DeepSeek Harness Reaches 1.54 Million Daily Users and Plans an Official Plugin Marketplace (6 November 2026, 01:50 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.