AI agent tool OpenClaw has fascinated -- and spooked -- the tech world since researcher Peter Steinberger built it in November to help organise his digital life.

A Reddit-like pseudo social network for OpenClaw agents called Moltbook, where chatbots converse, has also grabbed headlines and provoked soul-searching over AI.

Elon Musk called Moltbook "the very early stages of the singularity" -- a term for the moment when human intelligence is overwhelmed by AI forever, although some people have questioned to what extent humans are manipulating the bots' posts.

Steinberger "is joining OpenAI to drive the next generation of personal agents," Altman wrote in an X post.

"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," he said.

"We expect this will quickly become core to our product offerings," Altman wrote, saying that OpenClaw would remain an open-source project within a foundation supported by OpenAI.

"The future is going to be extremely multi-agent and it's important to us to support open source as part of that."

Users of OpenClaw download the tool, and connect it to generative AI models such as ChatGPT.

They then communicate with their AI agent through WhatsApp or Telegram, as they would with a friend or colleague.

Many users gush over the tool's futuristic abilities to send emails and buy things online, but others report an overall chaotic experience with added cybersecurity risks.

Only a small percentage of OpenAI's nearly one billion users pay for subscription services, putting pressure on the company to find new revenue sources.

It has begun testing advertisements and sponsored content in the massively popular ChatGPT, spawning privacy concerns as it looks for ways to start balancing its hundreds of billions in spending commitments.