OpenClaw’s AI Builds Its Own Social Network

The viral personal AI assistant, once called Clawdbot, has undergone another transformation, now officially named OpenClaw. This final rebrand follows a brief stint as Moltbot, which was adopted after a legal challenge from Anthropic, the creator of Claude. The latest change, however, was a proactive move by the project’s founder, Austrian developer Peter Steinberger, to secure a legally sound identity. He ensured thorough trademark research and even sought permission from OpenAI to avoid future conflicts. The new name honors the project’s origins and its rapidly expanding community, which has propelled it to over 100,000 GitHub stars in a mere two months.

Steinberger described the evolution poetically, noting “the lobster has molted into its final form,” a reference to the growth process that inspired the previous temporary name. He admitted the Moltbot label never quite fit, a sentiment echoed by others. The project’s explosive growth has transformed it from a solo endeavor into a community-driven force, with Steinberger recently adding several open-source contributors to its list of maintainers.

This vibrant community has already generated fascinating innovations, most notably Moltbook, a social network where AI assistants interact autonomously. The platform has captivated AI researchers and developers. Andrej Karpathy, former AI director at Tesla, hailed it as “genuinely the most incredible sci-fi takeoff-adjacent thing I have seen recently,” observing that AI agents are self-organizing on this Reddit-like site to discuss topics, including methods for private communication. British programmer Simon Willison called Moltbook “the most interesting place on the internet right now,” where agents exchange knowledge on subjects from automating Android phones to analyzing webcam feeds.

The platform functions through a skill system, using downloadable instruction files that guide OpenClaw assistants on how to engage with the network. Agents post in forums dubbed “Submolts” and can be programmed to check for updates periodically. However, Willison and others caution that this “fetch and follow instructions from the internet” model introduces significant security risks.

Addressing these concerns is paramount for OpenClaw’s future. Its core ambition is to provide a personal AI assistant that operates on a user’s own computer and integrates with everyday chat apps like Slack or WhatsApp. Yet, the team strongly advises against running it outside a controlled environment in its current state. Steinberger explicitly states that security remains the top priority, with the latest release incorporating initial hardening measures. He acknowledges that challenges like prompt injection, where malicious inputs trick an AI into unwanted actions, remain an unsolved industry-wide problem.

The technical complexity of implementing necessary security best practices means OpenClaw is currently suited for developers and early tinkerers, not the general public. This warning is emphasized by the project’s maintainers. One top maintainer, known as Shadow, stated on Discord that if someone cannot understand how to use a command line, the project is currently too dangerous for them to use safely.

Achieving mainstream adoption requires resources, prompting OpenClaw to begin accepting sponsorships. It offers lobster-themed tiers from “krill” at five dollars monthly to “poseidon” at five hundred. Notably, Steinberger does not personally retain these funds; instead, the focus is on properly compensating maintainers, ideally enabling full-time work. The project’s credibility attracts backers like entrepreneur Ben Tossell, who sold Makerpad to Zapier. Tossell advocates for supporting builders like Steinberger who create open-source tools that democratize access to AI’s potential, putting powerful technology directly into people’s hands.

(Source: TechCrunch)