AI Generated Image
OpenClaw 2.0 pours glitter on slow-burning security dumpster fire
The Register • • general
Key Points:
- OpenClaw has released its largest update yet, version 2.0, focusing primarily on usability improvements such as a simplified installation process and a redesigned browser interface resembling popular AI chat services.
- The update introduces shared cloud sessions, enabling multiple users to collaborate with a single AI agent while maintaining context, bringing OpenClaw closer to enterprise-level features offered by competitors like Anthropic and OpenAI.
- Despite these enhancements, security improvements remain limited; shared sessions lack strong isolation, secret credentials are not encrypted at rest, and sandboxing for untrusted code is disabled by default.
- OpenClaw has faced criticism for significant security vulnerabilities since its launch, including incidents of private data exposure and unauthorized actions by AI agents, raising concerns about the risks of widespread deployment without stronger protections.
- Overall, while OpenClaw 2.0 improves accessibility and user experience, it does not sufficiently address the platform’s critical security challenges, cautioning users against granting extensive system access to its AI agents.
