Baidu, Alibaba and Tencent promote AI agents in the face of the rise of OpenClaw in China

OpenClaw, which has spread rapidly in Chinese technology communities, is an open source program that connects language models with a computer to execute actions directly on the system, allowing it to perform tasks such as managing files, sending emails or using applications autonomously, based on user instructions.

LOOK: The new digital anxiety: 8 out of 10 people fear that someone will look at their phone screen, according to a study

Tencent announced this Wednesday that it is preparing the launch of ‘QClaw’, an assistant based on the OpenClaw open ecosystem that allows you to operate directly from applications such as WeChat – similar to WhatsApp, blocked in China – without the need for installation, in a bid to reduce barriers to use through its integration into services already widespread among users.

For its part, Baidu presented this Monday its range of agents under the ‘Longxia’ brand (‘lobster’, in reference to the OpenClaw logo), among which the ‘DuMate’ desktop assistant stands out, capable of automating various tasks, and ‘Xiaodu Longxia’, aimed at the home environment with voice control functions and execution of actions.

Company executive Shen Dou said, according to state television CCTV, that OpenClaw’s popularity points to an evolution toward ecosystems of specialized functions that will “fragment traditional software.”

In parallel, Alibaba reinforced its commitment to these systems with the launch this Tuesday of ‘Wukong’, an agent integrated into its DingTalk business platform – similar to Slack, blocked in China – and aimed at automating workflows within organizations, reported the Yicai media.

Security alerts

The rise of these systems – popularly known as ‘lobsters’ due to the aforementioned icon, which shows said crustacean – has been accompanied by security incidents, such as data leaks or improper access, which has led companies to try to offer more controlled environments.

In this context, the Ministry of State Security (MSS, China’s main intelligence agency) warned this week that these agents require elevated permissions that can facilitate remote access to devices or the exposure of sensitive information, in addition to warning of the risk of malicious use to spread false content or introduce harmful code through extensions.

Other organizations such as the Ministry of Industry and Technology have also noted that improper configuration can lead to “high cybersecurity risks,” such as data leaks or unauthorized access.

This trend is part of the rapid development of AI in China, where language models from companies such as Baidu, Alibaba, DeepSeek or Tencent proliferate, which have shown capabilities comparable to those of American systems, often at a lower cost, although doubts persist about the viability of their expansion due to the censorship exercised by the authorities.

According to a report published this week by the rating agency Moody’s, the revenue of large Chinese technology companies is registering growth driven by AI, although it warns that its integration increases operational, cybersecurity and data management risks.