OpenClaw Founder: Privacy Fully Embraced, but Security Concerns Still Loom

BlockBeats News, February 26th, OpenClaw founder Peter Steinberger said in a recent interview that OpenClaw has the most thorough privacy solution at the moment, but security is another matter. Currently, all data is stored locally, with no content uploaded to the cloud, giving users full control over access permissions and memory data, thus ensuring privacy.

However, the security risk lies not in being breached, but in being out of control. Peter stated that the security of the AI Agent largely depends on the capability of the underlying model. The Prompt Injection attack on large models is not impossible, but it is not as easy as people think. OpenClaw has now introduced dedicated security experts, and their current core work is to help users run securely in new scenarios as much as possible.

OpenClaw's security issues remain troubling. February's data shows that there are 341 malicious plugins in the skill marketplace, with a contamination rate of 11.3%, posing a serious supply chain risk. However, under the premise of "using the latest model + reasonable configuration," the self-defense capability of the AI Agent is stronger than what the outside world imagines.