144,000 OpenClaw instances are vulnerable to attack from the network


You can do practical things with OpenClaw – formerly Clawd and Moltbot. The software makes it possible to use various AI models as customizable helpers. Or letting them interact with other bots. The latter point in particular caused real public interest in the form of a Reddit replica for AI agents, called “Moltbook”. But the hype obscures two sober truths. First: Modern language models continue to be just probability-based word generators with no real intelligence. Second: Not only Moltbook, but also OpenClaw itself is a walking security disaster, especially in the hands of inexperienced users.



Source link