A series of four security flaws found in OpenClaw, a rapidly growing open-source platform for self-operating AI agents, has put about 245,000 public server instances at risk of being hacked, having credentials stolen, and being installed with secret backdoors. Originally started as “Clawdbot” in late 2025, OpenClaw links big language models to files, SaaS apps, … Continue reading OpenClaw Flaws Expose 245,000 Public AI Agent Servers