OpenClaw: The Viral AI Assistant That Lives on Your Device

What you need to know about this groundbreaking but security-challenged AI tool

Published on Feb. 4, 2026

Got story updates? Submit your updates here. ›

OpenClaw, an autonomous AI assistant that can run locally on users' devices and integrate with popular messaging apps, has gone viral in the tech community. While praised for its capabilities, the tool has also raised significant security concerns from experts who warn of potential risks from malicious code and data leaks. The project's evolution, from its initial "Warelay" incarnation to the current "OpenClaw" branding, has been marked by trademark issues and rebranding efforts.

Why it matters

OpenClaw represents a novel approach to AI assistants, allowing users to interact with a powerful AI agent directly through messaging apps rather than relying on centralized cloud-based services. This has sparked debates about the future of AI and whether decentralized, user-controlled models can challenge the dominance of tech giants in the AI assistant space. However, the security vulnerabilities identified in OpenClaw raise concerns about the risks of running such a powerful AI agent on personal devices.

The details

OpenClaw was originally created by Austrian developer Peter Steinberger as a weekend project called "Warelay," a bridge to let users interact with AI through messaging apps. The project later went through several rebrands, including "Clawdbot" and "Moltbot," before settling on the current "OpenClaw" name. Unlike centralized AI assistants like ChatGPT or Gemini, OpenClaw runs entirely on the user's own hardware, whether a laptop, home server, or cloud-based virtual private server. This allows the assistant to perform a wide range of tasks, from calendar management to home automation, by directly accessing the user's devices and applications. However, this also means OpenClaw requires administrative privileges, raising security concerns about the potential for malicious code or data leaks.

  • In late 2025, Peter Steinberger originally built the tool as a weekend experiment called "Warelay."
  • In early 2026, the project underwent its final rebrand to "OpenClaw" after previous names like "Clawdbot" and "Moltbot" faced trademark issues.
  • In late January 2026, Cisco's threat and security research team published an analysis warning that "from a security perspective, it's an absolute nightmare."

The players

Peter Steinberger

The Austrian developer who originally created the tool as a weekend experiment called "Warelay" in late 2025, before it went through several rebrands to become "OpenClaw".

Cisco

The technology company whose threat and security research team published an analysis in late January 2026 warning about the significant security concerns with OpenClaw.

Got photos? Submit your photos here. ›

What’s next

The creator, Peter Steinberger, has acknowledged that the project has grown beyond what he can maintain alone, raising questions about the long-term sustainability and security of OpenClaw.

The takeaway

OpenClaw's rise highlights the growing interest in decentralized, user-controlled AI assistants, but its significant security vulnerabilities underscore the challenges of developing such powerful tools without proper safeguards. As the AI landscape continues to evolve, the OpenClaw saga serves as a cautionary tale about the need to balance innovation with robust security measures.