WeChat has left a door open for AI phones.
Author | Huang Yu
WeChat is opening a controlled channel to mobile phone manufacturers’ AI assistants based on the A2A collaboration mechanism.
On June 4, it was reported that WeChat is collaborating with phone manufacturers such as Huawei, Honor, Xiaomi, OPPO, and vivo to launch A2A (Agent-to-Agent) assistant capabilities.
With this feature, users can directly initiate WeChat audio and video calls or send messages to friends using their phone’s voice assistant.
This capability does not mean WeChat is fully opening its ecosystem. Public information shows that this function allows manufacturers’ AI assistants to send commands to WeChat, which is responsible for execution and returning results. The process uses a dual authorization mechanism to ensure data security and privacy compliance.
Wallstreetcn has verified the authenticity of this news with multiple sources.
A person close to WeChat’s partners told Wallstreetcn that Honor had begun discussions with WeChat about A2A cooperation as early as 2025.
According to this source, the Honor Magic8 series, Honor 500 series, and the entire Honor X70 series are taking the lead in supporting WeChat’s A2A assistant capability.
After updating the AI assistant YOYO and WeChat to the latest versions on Honor phones, users can control WeChat via YOYO voice, sending messages or initiating audio and video calls.
WeChat has long taken a cautious stance towards phone AI assistant modes.
In 2024, phone manufacturers such as Huawei, Honor, OPPO, and vivo successively released AI phones and integrated AI capabilities into their systems, all touting the slogan: “Let your AI phone assistant help you with a single command.”
This cross-app capability was quickly resisted by existing software ecosystem giants.
Once it involves chat records, account security, payments, contacts, and user authorization, super Apps such as WeChat are reluctant to allow external Agents free access.
The Doubao phone at the end of 2025 is a typical example.
Bytedance, which pushed Doubao to become the top domestic AI-native application, successfully demonstrated the cross-APP operation capabilities of AI Agents through an "engineering prototype."
But soon, WeChat, Taobao, several bank apps, and others all set up “security firewalls,” launching countermeasures and a crackdown on the Doubao phone assistant.
At the end of 2025, multiple Doubao phone users found that if WeChat operations were involved, WeChat would crash abnormally or even become unable to log in.
A WeChat-related person told Wallstreetcn at the time: “It may have triggered existing security risk controls.”
Given conflicts of interest and data security concerns, it is clear that some third-party Apps are reluctant to grant phone AI assistants authorization.
This time, WeChat has not completely refused call access from phone Agents, but it has chosen a more controlled path: not allowing the phone assistant to bypass WeChat operations, but rather, within specific contexts, letting WeChat itself receive instructions, execute actions, and return results.
From the technical approach of phone manufacturers, compared with bypassing App authorization and relying on screen reading and simulated clicks by GUI Agents, obtaining App-side authorization via A2A or intent frameworks is becoming a safer path for implementation in high-frequency scenarios.
In 2025, vivo launched support for the A2A intelligent agent protocol. Zhou Wei, vivo Vice President, OS Product VP, and Head of vivo AI Global Research Institute, pointed out: “With this, developers can create agents and configure cards based on the Agent protocol fully online, then distribute them across all vivo channels using a unified intent framework and agent container.”
Similarly, for certain high-frequency scenarios and major internet firms, OPPO also prefers ecosystem interconnection via A2A.
WeChat’s decision to open some invocation capability at this time may also be related to the pace of developing its own WeChat AI agent.
On one hand, WeChat is developing its own AI Agent internally, while on the other, it is beginning to establish limited interoperability with third-party phone AI assistants via the A2A collaboration mechanism.
The former is a native agent within the WeChat ecosystem, the latter is controlled collaboration between WeChat and external phone Agents.
In 2025, Tencent management repeatedly mentioned the plan to build an AI Agent embedded in and operating based on WeChat’s unique ecosystem.
Since 2026, there have been continuous updates regarding this progress. Recently, the Financial Times reported that Tencent is accelerating the development of the embedded WeChat AI agent. The project is now in the prototype testing phase, with compliance procedures planned to start as early as this month, followed by testing with a small group of external users before gradually rolling out.
Tencent President Liu Chiping (Martin Lau) explained that this WeChat agent can leverage WeChat’s close connection with users and build a rich ecosystem, including mini program, content, transaction, social, and payment ecosystems, making the agent not just for chatting with users but able to perform many practical tasks.
However, there are significant concerns in launching this agent. Liu Chiping noted: “WeChat has a massive user base, faces privacy and security issues, and needs further improvement in model capabilities to address these.”
For WeChat, limited connectivity with phone AI assistants could also be a low-risk trial: observing user acceptance of voice-activated WeChat, and gaining experience for defining future invocation boundaries for its own agent.
Risk Warning and DisclaimerThe market involves risks, and investment should be cautious. This article does not constitute personal investment advice and has not taken into account the specific investment objectives, financial situation, or needs of any individual user. Users should consider whether any opinions, views, or conclusions in this article suit their specific circumstances. Investing accordingly is at your own risk.