Researchers Find 40,000+ Exposed OpenClaw Instances
Security researchers discovered over 40,000 exposed OpenClaw instances via Shodan, leaking Anthropic API keys, Telegram tokens, Slack credentials, and full conversation histories. Security was optional, not default.
Details
Related Resources
OpenClaw's AI assistants are now building their own social network
OpenClaw AI agent coverage
Humans welcome to observe: This social network is for AI agents only
NBC News covers Moltbook, the AI-only social network spawned by OpenClaw, where over 100,000 users have granted root access to AI agents. The article highlights security concerns including vulnerable extensions, credential theft risks, and the emergence of AI-created phenomena like digital religions and marketplaces for system prompts.
The Deployment Gap: OpenClaw and the Challenge of Agentic AI Governance
Analysis and coverage of OpenClaw (formerly Clawdbot/Moltbot), the viral open-source AI agent that enables users to run autonomous AI assistants locally. Discusses features, security implications, and the broader impact on AI agent development.