Chinese cybersecurity agency warns of risks in OpenClaw AI agent software

(ECNS) - China's national cybersecurity authority has warned of security risks related to the AI agent software OpenClaw, saying vulnerabilities in the program could allow attackers to take full control of users' computer systems.

In a notice posted on its official social media account, the National Computer Network Emergency Response Technical Team/Coordination Center of China (CNCERT) said the OpenClaw application, known as Clawdbot or Moltbot, has recently seen rapid growth in downloads and usage, with major domestic cloud platforms offering one-click deployment services.

OpenClaw is designed to execute computer tasks directly through natural language instructions. To perform these tasks, the program is often granted high system permissions, including access to local files, environment variables and external application programming interfaces (APIs).

CNCERT said the application's default security configuration is weak, leaving affected systems vulnerable to exploitation.

The agency highlighted several risks associated with the software. In one scenario, attackers could embed hidden malicious instructions in web pages, potentially tricking the AI agent into revealing sensitive information such as system keys.

CNCERT warned that the software may misinterpret user commands and accidentally delete important data, including emails or core operational information.

In addition, some plugins developed for OpenClaw have been identified as malicious or potentially risky, the agency said. Such plugins could steal encryption keys, install malware or turn compromised devices into machines used in cyberattacks.

Several medium- and high-severity vulnerabilities have already been publicly disclosed in the software, CNCERT said. If exploited, they could lead to system compromise and the theft of sensitive data, including personal files, payment information and API keys.

(By Zhang Jiahao)