March 11 (Reuters) - Chinese government agencies and state-owned enterprises have warned staff in recent days against installing artificial intelligence agent OpenClaw on office devices for security reasons, according to two sources familiar with the matter.
OpenClaw is an open-source software that can autonomously execute a wide range of tasks with minimal human guidance, going beyond the traditional research and query-answering capabilities of AI chatbots.
Over the past month, it has been enthusiastically
adopted and promoted by Chinese tech developers, leading AI companies, as well as several local governments based in China's tech and manufacturing hubs.
At the same time, central government regulators and state media have issued repeated warnings about OpenClaw's potential to inadvertently leak, delete, or misuse user data once downloaded and given security permissions to operate on a device.
The curbs suggest Beijing, while hoping to promote an "AI plus" action plan that aims to create innovation-driven growth by embedding the technology throughout the economy, is also wary of cyber and data security risks, amid intensifying geopolitical tensions.
One of the sources said staff at state-owned enterprises were told by regulators not to deploy OpenClaw, including in some cases on personal devices.
The second source, from a Chinese government agency, said the software had not been banned outright at their workplace but staff had been warned about safety risks and advised not to install it.
Both declined to be named as they were not authorised to speak to the media.
It is unclear how widespread the ban is and whether it will affect local government policies, which in some cases offer million-dollar subsidies to companies that innovate using OpenClaw. These policies have all been framed as local implementation of Beijing's national "AI plus" action plan.
A research centre under Chinese tech hub Shenzhen's municipal health commission last week ran an OpenClaw training session attended by thousands, as part of its "AI plus" push in healthcare.
It is also unclear whether the latest restrictions spell the end of all Chinese government deployment of OpenClaw. Shenzhen district Futian used OpenClaw to create an AI agent tailored to civil servant work, state-owned Southern Daily reported on Sunday.
China's state asset regulator and industry ministry did not immediately respond to a request for comment.
Bloomberg News first reported the restriction.
OpenClaw was developed by Peter Steinberger, an Austrian, and uploaded to GitHub last November. Steinberger was hired by OpenAI last month.
(Reporting by Beijing newsroom and Preetika Parashuraman in Bengaluru; Editing by Harikrishnan Nair and Kate Mayberry)
