China warns of security risks from popular AI agent OpenClaw

The Chinese Ministry of State Security has issued a warning regarding security risks associated with the open-source AI agent OpenClaw, which has recently become a phenomenon among Chinese developers and users. The ministry noted that users often grant the system high-level permissions to complete tasks, which could allow attackers to gain remote control of devices or access sensitive information. The agency advised users to limit permissions, verify the origin of add-ons, and run the system in isolated environments to mitigate cybersecurity threats. Created by Austrian developer Peter Steinberger, OpenClaw has seen rapid adoption in China, with over 23,000 active instances reported in the country.