China’s government agencies and state-owned enterprises have begun warning staff against installing the artificial intelligence agent OpenClaw on office devices, citing potential security concerns.
The guidance, as reported by Reuters, highlights caution inside the Chinese state sector over emerging autonomous AI tools.
OpenClaw is an open-source system capable of independently performing complex digital tasks with limited human guidance.
The tool has gained traction among Chinese developers, AI companies, and several local governments during the past month.
Regulators and state media have warned that once installed and granted system permissions, the software could leak, delete, or misuse sensitive data stored on devices.
The warnings show the balance Beijing is attempting to maintain as it promotes artificial intelligence adoption through its national “AI plus” strategy while guarding against cyber and data risks.
AI agent security concerns
The concerns centre on OpenClaw’s ability to autonomously carry out actions on a computer once given access rights.
Unlike conventional AI chatbots that mainly generate text or answer questions, AI agents like OpenClaw can execute tasks automatically.
These may include interacting with files, launching programs, or navigating online services without direct supervision.
As per the report, Chinese regulators have signalled that such capabilities could create security vulnerabilities if used within government or enterprise systems.
Officials and state media have warned that AI agents with elevated permissions may expose confidential information or interfere with sensitive digital infrastructure.
Staff working at state entities have therefore been advised to avoid installing the software on work computers.
Warnings to state workers
As per the report, regulators instructed employees at several state-owned enterprises not to deploy OpenClaw.
In some cases, staff were also discouraged from installing the software on personal devices used for work-related activities.
Another source cited by the report, said OpenClaw had not been formally banned within their organisation.
However, employees were warned about safety risks and advised not to install it.
It remains unclear how widely the restrictions have been implemented across government bodies and state companies.
Local government support grows
Despite the warnings from regulators, OpenClaw has been promoted across China’s technology ecosystem during the past month.
Developers, AI firms, and several local governments have encouraged experimentation with the software.
Some regional authorities have offered million-dollar subsidies to companies developing applications based on OpenClaw.
These incentives are framed as part of local implementation of Beijing’s broader “AI plus” action plan to integrate artificial intelligence across the economy.
The push has extended into healthcare. Last week, a research centre under Shenzhen’s municipal health commission hosted a training event focused on OpenClaw.
Thousands attended the session as part of the city’s effort to accelerate AI adoption in medical services.
The post China warns state staff over OpenClaw AI agent security risks appeared first on Invezz