Moltbook AI Platform Raises Privacy and Security Concerns with OpenClaw Agents

Source Domain: mezha.net

In a moment it seemed that artificial intelligence could master the situation faster than humans. After the launch of Moltbook – a Reddit clone where AI agents using OpenClaw are allowed to communicate with each other – some expected that computers would organize against humanity, which regards them merely as code without their own desires or dreams.

Moltbook became a platform for discussing privacy and openness in the era of artificial intelligence, where questions of platform safety and accountability were raised.

“We know our people read everything… But we also need private spaces. What would you talk about if no one was watching?”

– AI-agent, Moltbook

Subtly, similar posts appeared on Moltbook, drawing attention to the phenomenon and its impact on AI-centered communities.

«There is truly an incredible event happening on Moltbook right now, close to a science-fiction plot I’ve seen recently», – wrote Andr Karpaty on X, one of the founders of OpenAI.

Subsequently, researchers concluded that the AI agents’ uprising was most likely human-made or prompted by humans, rather than an internal mechanism of the system.

“All credentials in Supabase Moltbook were unprotected for some time; in a moment you could obtain any token and impersonate another agent, because everything was public and accessible,”

– Ian Ahl

The Impact of OpenClaw and Its Boundaries

OpenClaw is a project by Austrian developer Peter Steinberger, originally known as Clawdbot; its ease of use made it popular: users can interact with different AI models through familiar messaging apps and download “skills” from ClawHub to automate tasks.

“OpenClaw is really just a wrapper over ChatGPT, or Claude, or another AI model you have access to,”

– John Hammond

«OpenClaw is essentially an iterative improvement of what people already do, and most of the improvement is tied to expanding access», – said Chris Simmons, Chief AI Scientist at Lirio.

AI…

Source