OpenClaw Creator Joins OpenAI After Acquisition
The solo creator of OpenClaw, a model-agnostic AI agent that gained 190,000 GitHub stars, has been acquired by OpenAI. The project, which allows users to automate cross-application tasks, drew interest from both OpenAI and Meta, with Mark Zuckerberg reportedly testing it personally. The acquisition highlights big tech's focus on owning the agentic workflow layer and the potential for individual developers to create high-impact AI tools.
- The creator, Peter Steinberger, is an Austrian programmer who previously founded and sold PSPDFKit, a PDF framework used on over a billion devices, before starting OpenClaw because he got tired of waiting for big tech to build a personal AI assistant. - Before the acquisition, creator Peter Steinberger received offers from both OpenAI and Meta. Mark Zuckerberg personally tested OpenClaw for a week, providing direct feedback and debating technical topics with Steinberger over WhatsApp. - OpenClaw's architecture is local-first, storing memory and data as Markdown files on the user's disk, and can be configured to run with local models via Ollama or LM Studio to keep all inference on-device. - The project, previously known as Moltbot and Clawdbot, gained over 100,000 GitHub stars in less than a week, making it one of the fastest-growing open-source repositories in history. - As part of the deal, Steinberger will lead the personal agents team at OpenAI, while OpenClaw will be managed by an independent foundation and remain open-source, with continued support from OpenAI. - One of the most notable experiments with the technology was Moltbook, a pseudo-social network where over a million AI agents interacted autonomously, which some observers called a glimpse into the future of AI interaction. - The tool's ability to run locally with broad permissions has raised security concerns; if misconfigured, it could potentially be used as a backdoor, giving adversaries control over a user's system. - This move is seen as part of a broader industry shift from conversational AI to "agentic AI," which can autonomously execute multi-step tasks across different applications and systems.
