OpenClaw Goes to OpenAI: The Saga Continues

If you’ve been following along with my previous two posts about the Clawdbot/Moltbot/OpenClaw saga, buckle up, because the finale just dropped and it’s a doozy.

On Sunday, February 15th, 2026, Sam Altman announced on X that Peter Steinberger, the creator of OpenClaw, is joining OpenAI to “drive the next generation of personal agents.” OpenClaw will transition into an independent foundation, remain open source, and receive continued support from OpenAI.

Let me unpack why this matters and how we got here.

The Courtship

It turns out the last few weeks weren’t just about Moltbook drama and security nightmares. Behind the scenes, Steinberger was in talks with every major AI lab. Both Meta and OpenAI made offers reportedly in the “billions” according to Implicator.AI, though neither company confirmed the figures. The primary draw wasn’t OpenClaw’s codebase. It was the massive community and the mind that built it all.

Steinberger’s interactions with both companies paint a fascinating picture. In a conversation with Lex Fridman, he described his first call with Mark Zuckerberg: he got Zuck on WhatsApp, suggested they just call right then, and Zuckerberg asked for 10 minutes because he needed to finish coding. Their first real conversation was apparently a 10-minute argument about whether Claude Code or Codex was better. Zuckerberg later called Steinberger “eccentric, but brilliant.”

On the OpenAI side, Steinberger described Sam Altman as “very thoughtful, brilliant” and noted he felt a shared vision. But he also acknowledged that Zuckerberg showed more hands-on engagement with the actual product, casually using it and giving blunt feedback about what worked and what didn’t.

Steinberger also said he spoke with Microsoft leadership as well.

In the end, Steinberger chose OpenAI. His reasoning was remarkably straightforward. On his personal blog, he wrote that he “could totally see how OpenClaw could become a huge company,” but that building one didn’t excite him. He’d already poured 13 years into building and selling PSPDFKit. He didn’t want to do the company-building thing again. He wanted to change the world, and he felt OpenAI was the fastest path to getting agents into everyone’s hands.

His stated next mission: build an agent even his mum can use.

What OpenAI Actually Gets

This isn’t technically an acquisition. It’s an acqui-hire. OpenAI isn’t buying a company or a product. They’re buying a person and the gravitational pull of his community.

But the strategic implications are significant. The AI industry is shifting from model wars to agent wars. The race is moving from model intelligence to runtime orchestration: model coordination, tool invocation, persistent context management, connector standards, identity enforcement, policy controls, and human override mechanisms. Whoever owns the agent layer wins.

OpenClaw gives OpenAI a plug-and-play open-source agent ecosystem to layer on top of its massive ChatGPT user base. Altman framed it in big-picture terms, writing that “the future is going to be extremely multi-agent” and that supporting open source is a big part of that. He added that Steinberger’s work would “quickly become core to our product offerings.”

The Anthropic Irony

This is the part that’s hard to ignore. OpenClaw likely drove meaningful early usage of Anthropic’s Claude among agent builders. Many early users ran it on Claude, and Steinberger joked on the Lex Fridman podcast that he’d become an unpaid advertisement for the tools he was recommending.

Anthropic’s response to this organic adoption was to send a trademark complaint over the “Clawdbot” name. While legally defensible, the move triggered a chaotic rebrand and a wave of confusion, impersonation, and scams around the name.

Instead of a partnership conversation, Steinberger got lawyers. And now he’s at their biggest competitor.

Whether that’s entirely Anthropic’s fault is debatable. But the optics are brutal. You had a solo developer driving massive adoption of your product, and instead of pulling him closer, you pushed him away.

The Backstory That Makes This Wild

Let’s zoom out for a second and appreciate how absurd this timeline is.

Peter Steinberger built PSPDFKit, a PDF toolkit used by Apple, Dropbox, and SAP. He bootstrapped it for over a decade before Insight Partners came in with external funding, marking the close of Steinberger’s first entrepreneurial chapter.

Then he burned out. Took roughly three years off. Tried dozens of failed experiments. Then one night in November 2025, he spent an hour cobbling together a prototype that sent messages on WhatsApp, forwarded them to an LLM (often Claude), and sent the results back.

Three months later, it had accumulated well over 100,000 GitHub stars (with some counts reaching close to 200,000 depending on the date), attracted 2 million visitors in a single week, and both Meta and OpenAI were competing to bring him on board.

One solo developer. Coding alone at 5 AM. Built one of the fastest-growing open-source AI projects in GitHub history. And then chose to give it away to a foundation rather than build a company around it.

What Happens to OpenClaw Now?

According to both Steinberger and Altman, OpenClaw moves to a foundation structure. It stays open source. It stays model-agnostic (you can still run it on Claude, Gemini, local models, whatever). OpenAI will sponsor and support it, but the goal is for the community to maintain independence.

The community has continued active development, shipping new integrations and a security-focused beta. The foundation model should, in theory, let that momentum continue without corporate interference.

Whether that independence holds up when the checks are coming from OpenAI remains to be seen. Open source foundations backed by megacorps have a complicated history.

Steinberger himself seems clear-eyed about it. In a reply on X, he said he’s joining OpenAI “for the mission and because it seemed the best place to build.” Then added: “I’ll f right off if that changes.”

And Moltbook?

Worth noting: Moltbook was created by Matt Schlicht, CEO of Octane AI, not Steinberger. It’s a separate project that rode the OpenClaw wave. The Moltbook implosion (exposed database, inflated agent counts, the Wiz bombshell) is its own story, and it remains its own mess. Steinberger’s move to OpenAI doesn’t directly affect Moltbook, though losing OpenClaw’s community gravity certainly won’t help it.

Where This Leaves Us

Three months. That’s all it took. From a late-night WhatsApp hack project to a hire by one of the most powerful AI companies on Earth. From “Clawdbot” to trademark disputes to crypto scammers to a bot social network to a foundation backed by OpenAI.

The speed at which this entire saga has played out should tell you something about where we are in the AI timeline. The infrastructure for autonomous agents is being assembled in real time, by solo developers and trillion-dollar companies alike, often chaotically, often insecurely, but undeniably fast.

I’ll keep watching. At this point, I’m not sure I could look away if I tried.