OpenClaw Security Testing: 80% hijacking success on a hardened AI agent(earlycore.dev)
5 pointsby paldeac27 hours ago1 comment
  • paldeac27 hours ago
    Would love to see other vulnerabilities found