The open-source AI tool OpenClaw has attracted a lot of attention from users around the world in recent weeks. The software, developed by Austrian engineer Peter Steinberger, allows users to automatically perform various tasks on the computer through an AI assistant.
Thanks to its open source code, OpenClaw can be freely downloaded and customized, which has contributed to its rapid popularity. However, cyber security experts warn that this very openness can pose a serious security risk.
OpenClaw has extensive computer access powers
Unlike AI tools such as: ChatGPT or Gemini that function as web or mobile applications, OpenClaw is installed directly into the computer system and is given broad powers to manage the device.
This means that AI can open documents, send messages, manage e-mails, view files or even write programming code based on commands sent by the user through communication applications.
It is precisely because of this approach that experts warn that installing OpenClaw practically means giving full access to the system.
The security team of XinDun Security warns that OpenClaw can see everything that appears on the computer screen. If the user downloads a modified or malicious version of the software, there is a risk that attackers will get access to sensitive information such as passwords for online banking, private messages or business data.
An additional problem is the fact that the software is installed deep into the operating system in order to obtain the necessary authority to control the device. Therefore, its complete removal can be complicated, especially for users who do not have advanced technical knowledge.
Experts warn that in such cases, OpenClaw can leave parts of the code in the system that can potentially serve as a “backdoor” for hackers, which in the long run represents a serious security risk for users.