OpenAI Hires OpenClaw Creator to Lead Personal AI Agents

OpenAI has hired Peter Steinberger, the Austrian developer behind the viral open-source agent OpenClaw, to lead the company’s next generation of personal AI agents. CEO Sam Altman announced the hire on Saturday, calling Steinberger “a genius with a lot of amazing ideas about the future of very smart agents interacting with each other to do very useful things for people.”

The move signals OpenAI’s intensifying push into autonomous AI agents—software that doesn’t just answer questions but takes actions on behalf of users. OpenClaw, which Steinberger built as a solo developer over the past few months, grew rapidly after going viral in late January.

“My next mission is to build an agent that even my mum can use,” Steinberger wrote in a blog post explaining his decision. “I’m a builder at heart. What I want is to change the world, not build a large company, and teaming up with OpenAI is the fastest way to bring this to everyone.”

From PSPDFKit to OpenClaw

Steinberger spent 13 years building PSPDFKit, a developer tools company focused on the Apple ecosystem. After a three-year break, he returned to software development—but in an entirely different field. OpenClaw, which cycled through the names Clawdbot and Moltbot before settling on its current identity, is a TypeScript-based web application built by someone who had spent two decades writing iOS and macOS code. The pivot itself is a testament to how AI tools have lowered the barrier to building in unfamiliar domains.

The project went viral in late January 2026, credited partly to its open-source nature. OpenClaw’s appeal was straightforward: an agent that runs locally, executes shell commands, manages files, and connects to external services through a growing library of community-built skills.

That rapid adoption came with growing pains. Independent security researchers found skills on OpenClaw’s ClawHub marketplace contained malicious code—a reminder that the extensibility making agent platforms useful also makes them vulnerable. Steinberger moved quickly, partnering with VirusTotal to implement automated scanning, but the incident highlighted structural challenges facing any open agent ecosystem.

OpenClaw Moves to a Foundation

Rather than absorbing OpenClaw into its corporate structure, OpenAI is sponsoring the project’s transition to an independent open-source foundation. Altman framed the decision in multi-agent terms: “The future is going to be extremely multi-agent, and it’s important to us to support open source as part of that.”

The OpenClaw repository will continue accepting community contributions, and the foundation structure is designed to prevent any single company from controlling its direction. OpenAI’s sponsorship ensures continued development while preserving the project’s independence—a model similar to how major tech companies support the Linux Foundation or the Apache Software Foundation.

The arrangement also reflects a broader industry shift. The skills framework pioneered by Anthropic and now adopted by OpenAI has become a de facto standard for extending AI agents with specialized capabilities. By keeping OpenClaw open, OpenAI ensures compatibility with this emerging ecosystem rather than fragmenting it.

What This Means for OpenAI’s Agent Strategy

Steinberger’s hire fits into a pattern of aggressive agent-focused moves at OpenAI. The company launched ChatGPT Atlas with agent mode last October, giving its browser the ability to perform multi-step tasks autonomously. Earlier this month, it released enterprise agent management tools alongside its latest model updates.

But building agents that work reliably for mainstream users—Steinberger’s stated mission—remains an unsolved problem. Current AI agents excel in demos and controlled environments but struggle with the unpredictable complexity of real-world tasks. Getting an agent to book a flight is impressive; getting it to handle the edge cases when that flight is cancelled requires a different level of robustness.

Steinberger’s experience building a product that tens of thousands of people actually used daily gives him practical insight that pure research hires lack. He understands not just what agents can do theoretically, but what breaks when real users push them in unexpected directions.

Altman said the work would “quickly become core to our product offerings”—language that suggests OpenClaw’s architecture and lessons could influence how ChatGPT itself handles agentic tasks. For OpenAI, the hire is a bet that the person who built the most popular open-source agent in a matter of weeks can do something even more ambitious with access to the company’s frontier models and infrastructure.

For the broader agent ecosystem, the acquisition-by-hire raises a familiar tension. Open-source projects that attract corporate sponsors gain resources but risk losing community trust. Whether OpenClaw’s foundation model preserves genuine independence—or becomes a formality—will depend on how OpenAI handles governance in the months ahead.