OpenClaw, or MoltBot, or Clawdbot, whatever it's called this week, is the best thing to happen to Al security this year.
Security vendors are calling it a crisis, but this leaky AI project might save us all.
OpenClaw, a viral open-source AI agent, has caused a security firestorm, with major vendors like Cisco and Palo Alto labeling it a nightmare. Despite racking up security incidents and 400 malicious marketplace skills in weeks, its 2 million GitHub visitors and 180,000 researchers are stress-testing it openly. This public experimentation is exposing critical architectural vulnerabilities in agentic AI—like instant hijacking via malicious links—providing real-world threat models that closed enterprise platforms hide behind SOC 2 badges.
Why It Matters
The public flaws in this open project are forcing the entire industry to build real security for AI agents before they're deployed everywhere.