20 Organizations Publish AI Agent Self-Regulatory Convention
China Academy of Information and Communications Technology and 19 other organizations jointly drafted and published the Agent Development Self-Regulatory Convention at the Wuzhen Summit, according to IT之家 and 36氪. Its eight measures cover agent trust, interconnected infrastructure, interoperability, security, governance, user rights, industry deployment, and shared access. The convention proposes measures including verifiable identity mechanisms, tiered authorization, rule embedding, compliance testing, operational monitoring, and protocol compatibility and mutual recognition. It is a voluntary commitment without external enforcement, penalties, or an exit mechanism. Separately, The Paper reported that 155 applications had refused screen-automation access for Doubao Phone Assistant; the reports do not show that the convention changed those platform access policies.
The convention provides a common framework for agent identity, authorization, monitoring, and cross-platform interoperability, which could reduce the cost of adapting agents to different services. Its practical significance depends on whether participants implement these measures in specific products; signing it does not automatically make platforms open their services.
华为、阿里、腾讯、小米、百度、火山引擎等 20 家单位共同起草发布《智能体发展自律公约》,抵制利用智能体实施算法压榨、黑灰产等行为 - IT之家
- 日夜间 随系统 浅色 深色
- 主题色 黑色
#China Academy of Information and Communications Technology#Huawei#Alibaba#Tencent#Xiaomi#Baidu#Volcano Engine#Ant#Douyin