AI Security · Web App Attack

OpenClaw Flaw Lets Websites Hijack Local AI Agents

Successful attack granted admin access, enabling configuration and log disclosure and node enumeration.

5 sources · Mar 6

Timeline

• Mar 6 Reported by BleepingComputer
• Mar 3 Reported by Dark Reading
• Feb 28 Reported by The Hacker News

Sources

• Mar 6 BleepingComputer

  Bing AI promoted fake OpenClaw GitHub repo pushing info-stealing malware

  Fake OpenClaw installers hosted in GitHub repositories and promoted by Microsoft Bing's AI-enhanced search feature instructed users to run commands that deployed information stealers and proxy malware.

  original

• Mar 3 Dark Reading

  Critical OpenClaw Vulnerability Exposes AI Agent Risks

  The now-patched flaw is the latest in a growing string of security issues with the viral AI tool, which has seen rapid adoption among developers.

  original

• Mar 2 SecurityWeek

  OpenClaw Vulnerability Allowed Websites to Hijack AI Agents

  Malicious websites could open a WebSocket connection to localhost on the OpenClaw gateway port, brute force passwords, and take control of the agent.

  original

Part of the PlainSec briefing for 2026-03-01

Every edition of this story: OpenClaw Flaw Lets Websites Hijack Local AI Agents