Raising lobsters requires a tight safety string! OpenClaw still carries risks after updates

Under the current trend, security risks cannot be ignored - the Ministry of Industry and Information Technology has previously issued a special warning reminder. On the evening of March 10th, experts once again issued a statement reminding that even if the "lobster" intelligent agent has been updated to the latest version, it does not mean that security risks have been completely eliminated. "Lobster farming" needs to always tighten the string of security protection.

OpenClaw is called the "lobster" because of its red lobster shaped icon, formerly known as Clawdbot or Moltbot. It is an open-source AI agent with persistent memory and active execution capabilities, which can be privately deployed locally. By integrating communication software and large language models, it achieves autonomous implementation of user instructions. Its convenience and efficiency have quickly gained popularity, but it is precisely this characteristic of "autonomous decision-making and calling system resources" that has planted many security risks and become a key warning object for the Ministry of Industry and Information Technology and experts.

As early as February 5th, the Network Security Threat and Vulnerability Information Sharing Platform of the Ministry of Industry and Information Technology had issued a warning on preventing security risks of OpenClaw open source AI intelligent agents. It clearly pointed out that the agent has high security risks in default or improper configuration, which can easily lead to network attacks, information leaks, and other problems. At the same time, targeted prevention suggestions were given to sound the alarm for users. On the evening of March 10th, Wei Liang, Vice President of the China Academy of Information and Communications Technology, and other experts further added reminders, breaking the misconception that "version updates are security" - although the latest version can fix known security vulnerabilities, it cannot completely eliminate all risk hazards.

Experts explain that the security risks of OpenClaw stem from the dual combination of its own characteristics and ecological shortcomings. On the one hand, in order to achieve the ability to autonomously execute tasks, it is granted high system permissions by default, which can access the local file system, read environment variables, call external APIs, and install extension functions. However, its default security configuration is extremely fragile, and once an attacker finds a breakthrough, it is easy to obtain complete control of the system. On the other hand, the current skill pack market for "Lobster" lacks strict review, and some plugins on community platforms such as ClawHub pose a risk of malicious poisoning. After installation, it may lead to key theft, Trojan implantation, and other issues. Even if upgraded to the latest version, without targeted protection, it may still be attacked.

Specifically, the security risks that OpenClaw has revealed mainly focus on four aspects: first, the risk of "prompt word injection", where attackers can hide malicious instructions through web pages and induce agents to leak user system keys; The second is the risk of "misoperation", as misunderstandings of user instructions may result in the accidental deletion of important information such as emails and core production data; The third risk is the poisoning of functional plugins, as malicious plugins can lead to device control and sensitive data leakage; The fourth is the risk of security vulnerabilities. Currently, multiple high and medium risk vulnerabilities have been publicly exposed, which may lead to serious consequences such as system control and privacy breaches. For key industries, it may even cause business system paralysis.