OpenClaw’s creator has joined OpenAI — here’s why that’s a big deal

OpenAI CEO Sam Altman has announced that Peter Steinberger, the creator of AI assistant OpenClaw, is joining OpenAI to “help drive the next generation of personal agents.” Altman expects OpenClaw’s capabilities to “quickly become core” to OpenAI’s products.

What is OpenClaw?

The AI with… hands?

OpenClaw is a self-hosted AI agent that’s capable of performing tasks like managing emails and automating workflows. In contrast to chatbots like ChatGPT, OpenClaw can actually “do things.” Steinberger has described traditional AI as a “brain in a jar” and OpenClaw as “hands” that allow the brain to interact with the world. In other words, instead of just answering prompts, it can actually perform real-world tasks. It’s either super cool or super creepy, depending on who you ask.

The OpenClaw timeline

Practically an overnight sensation

Although the AI agent has gone viral over the last few weeks, things have moved so quickly that it’s been hard to keep up. Here’s the basic breakdown:

• OpenClaw was launched by Peter Steinberger in November of last year as a small project with the goal of using Anthropic’s Claude AI to actually accomplish tasks, rather than just provide info or other outputs. It was originally called Clawdbot (get it?). The project began to gain a cult following on GitHub.
• In late January, a major update was released that essentially changed OpenClaw from a simple script into a proactive AI agent. It gained the ability to read what was on your computer, use the mouse and keyboard, and take screenshots, making it much more capable. After this update, the tool’s popularity began to explode.
• Also in late January, Anthropic used legal pressure to force Clawdbot to change its name. Steinberger initially went with Moltbot before eventually settling on OpenClaw.
• Moltbook was launched on January 28th. Moltbook is a social network for AI agents, where OpenClaw-powered bots post back and forth to each other (yes, it’s as weird as it sounds). This quickly achieved meme status and helped propel OpenClaw out of enthusiast, power-user circles and into the mainstream.
• In early February, demand for Mac Minis to run OpenClaw caused noticeable supply issues. The Mac Minis were being used as dedicated “servers” to run OpenClaw, since it needs to run 24/7 for maximum effect.

And that brings us to the present. On February 15th, Altman announced that Steinberger was joining OpenAI (hopefully he’s more helpful than the ads the company is introducing to ChatGPT).

Is this really the future of AI?

Maybe

Steinberger joining OpenAI could actually be a pretty big deal, for a couple reasons. First, if the future of AI is in personal agents, hiring Steinberger could give OpenAI a significant strategic advantage over rivals like Meta and Anthropic. Meta was evidently also trying to hire Steinberger. Meanwhile, Anthropic totally blew it — the company seemed more worried about the project’s name than what it was capable of.

OpenAI’s support should also bring significant resources to help OpenClaw advance and to help patch some of its weak spots, like security (in order to function properly, OpenClaw really needs access to your entire device, which has earned it the title “security nightmare”). OpenAI’s resources and expertise should help shore things up in this regard.

Finally, the acquisition is also kind of a big win for open source software. Rather than being absorbed into ChatGPT, OpenClaw will continue to live on in an independent foundation with OpenAI support. This runs counter to the typical pattern of open-source projects being bought and killed by big tech.

Time will tell whether this is just more AI hype or the start of something big. After all, not everything OpenAI tries is a huge success. What do you think? Is this exciting news, or are you over it? Let us know.