How do major apps view AI agents controlling user access?

Question

Accepted Answer

Major apps restrict AI agents because they want to keep direct control of user access

Major apps view AI agents that can take over tasks across apps as a problem because it can shift control away from the platform they own. When firms such as Alibaba, Tencent, and JD.com restricted the built-in assistant’s access to their platforms, the AI agent on the Doubao-powered phone was effectively disabled, and ByteDance later turned off the tool in certain circumstances including when payments were involved. Industry commentary described this as platforms trying to avoid “losing control to another party” by keeping direct contact with users.
Platforms say direct user contact is tied to control
Platforms want to maintain direct contact with their users, because otherwise, they risk “losing control to another party.” This framing was highlighted by Kiranjeet Kaur, associate research director at IDC, in the context of AI agent devices that seek broad access to apps. The issue is not only technical access, but also who controls the user experience across services.
Access restrictions can neutralise an AI agent’s cross-app functions
After the Doubao phone prototype was released with the ability to execute tasks across apps using voice commands, major tech firms restricted the built-in assistant’s access to their platforms. That move effectively disabled the phone’s AI agent, demonstrating how platform controls can limit what an agentic smartphone can do in practice. ByteDance then turned off the tool in certain circumstances, including payments, after those blocks.
Payments and transactions are treated as a sensitive access boundary
The restrictions extended to scenarios involving money movement, since ByteDance disabled the tool when payments were involved. This indicates that even when an AI assistant can operate in some areas, transactional access is more tightly controlled by platform owners. The result, as described, was a narrower set of agent actions once platform access was restricted.
The race is unresolved, with performance still patchy
Even with rapid showcases of agentic phones at events like the World Artificial Intelligence Conference in Shanghai, analysts said there is “no clear winner.” Marc Einstein of Counterpoint Research said AI agent tools are still often patchy, which helps explain why platforms may keep tight control over access. He also predicted that in five to 10 years, phones may not use apps the way people do today, implying the current model is still in flux.

Reading history

Hide →

This history is cleared when the page is closed.

How do major apps view AI agents controlling user access?

Loading

Was this article helpful?

Helpful

Not helpful

Report an issue

1

0

0

Share with friends