Sam Altman says Peter Steinberger, founder of OpenClaw, is joining OpenAI “to drive the next generation of personal agents”; OpenClaw will remain open source
Peter Steinberger is joining OpenAI to drive the next generation of personal agents. He is 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. We expect this will quickly become core to our
@samaSam Altman
Context & Ripple Effects
OpenClaw had only recently emerged from a forced rebranding sequence after Anthropic objected to an earlier name, making its creator's move a notable turn for a project still positioned as open source. The arrangement separates the project's availability from its founder's new institutional home.
Related coverage also reports that Meta offered Steinberger more money, but that he chose OpenAI's vision, underscoring that personal-agent expertise has become a contested recruiting target among frontier labs.
First-order effects
OpenAI gains Steinberger's agent-design expertise for its personal-agent work, while OpenClaw remains available as an open-source project rather than being folded into a proprietary product.
OpenClaw users and contributors retain access to the project despite its founder joining OpenAI, limiting immediate disruption to its existing community.
Second-order effects
The reported competition with Meta for Steinberger raises the stakes for other AI labs recruiting builders with proven agent projects, not just model researchers.
Keeping OpenClaw open source lets its ecosystem continue to develop independently while OpenAI can draw on its founder's experience, creating a closer link between open-source experimentation and lab product strategy.
Third-order effects
If similar hires continue, successful open-source agent projects may increasingly serve as talent pipelines into major AI labs, even when the underlying code remains public.
The episode points toward competition over the product layer around models: labs will differentiate through agents that can act for users, alongside the economics and trust constraints of deploying them.
The trend: Frontier AI labs are competing to turn capable models into personal agents, with open-source agent communities becoming an important source of product talent and ideas.
I'm joining @OpenAI to bring agents to everyone. @OpenClaw is becoming a foundation: open, independent, and just getting started.🦞 https://steipete.me/...
meta has absolutely fuck all except maybe manus. everything else is chaos and halfbaked. they also have no model. openai's been trailing anthropic in product, with many half hearted attempts to close up. with codex, they've created a good cc competitor.
pretty interesting to see all those “anthropic fumbled” posts. anthropic's been the only lab with a pretty coherent product strategy. can argue about execution, but at least they have a plan. that includes preexisting products competing with openclaw. they don't need peter.
All right @sama congratulations for getting @steipete and having him help you build out Codex. Great news for all and even better he is not being Zucked. clawON
I do wonder if Anthropic regrets not courting @steipete more aggressively to get him on Team Claude. On the other hand, this could have been purposeful since (per Pete's interview on Lex) OpenClaw is deeply unprofitable. Interesting to see how this plays out.
Excited to work with Peter Steinberger to build the future of agents for everyone and to continue to improve Codex in leaps and bounds. We are committed to OSS, continuing to make OpenClaw flourish and bringing agents to life in a way that is fun, safe and highly productive.
Feels like OpenClaw will be acquired by OpenAI. Events around this has been quite interesting: > @steipete calls his open source agent operating system Clawd > works on it day and night for it > clear tribute to Claude > Clawd gets popular > Claude strikes with lawyers > Peter