China has issued directives to state-run enterprises and government agencies, including major banks, to restrict the installation and use of OpenClaw AI software on office computers. This move comes amid growing concerns over potential security risks associated with the autonomous AI agent, which requires broad access to private data and can communicate externally.
Authorities have warned government agencies and state-owned enterprises, including the nation’s largest banks, against installing OpenClaw AI applications on office devices. This swift action aims to mitigate potential security risks that have emerged as companies and consumers across China have begun experimenting with the agentic AI phenomenon. Some entities were instructed to notify superiors if they had already installed the software, initiating security checks and potential removal processes.
The restrictions extend to certain employees within state-run banks and government agencies, prohibiting the installation of OpenClaw on both office computers and personal phones utilized on company networks. In some instances, the ban has reportedly been extended to the families of military personnel. While some notices mandate an outright ban, others require prior approval before the software can be used, suggesting a varied enforcement approach.
OpenClaw, previously known as Clawdbot and Moltbot, has gained popularity for its autonomous capabilities, such as managing inboxes, making reservations, and checking in for flights. Developed by Austrian developer Peter Steinberger, it can connect to messaging apps like WhatsApp and Slack. However, its ability to operate autonomously, access private data, and communicate externally has raised significant security alarms. Cybersecurity experts have warned that this combination of features creates a “lethal trifecta,” potentially exposing systems to external attacks and data leaks. Security researchers have also identified critical vulnerabilities in OpenClaw, including those allowing for gateway compromise and authentication token theft.
The news of the restrictions has impacted Chinese AI and tech stocks, with companies like Tencent Holdings Ltd. seeing their gains diminish. This development highlights a contradiction, as some local governments, such as those in Shenzhen and Wuxi, are simultaneously offering substantial subsidies for companies developing on the OpenClaw platform. This creates a complex landscape where national security concerns clash with local economic incentives for AI development.