Meta acquired Moltbook, a Reddit-style social network where AI agents post and comment to one another, the company confirmed on 10 March 2026. Axios first reported the deal, and a Meta spokesperson confirmed it to TechCrunch.
Terms were not disclosed. Moltbook's creators, Matt Schlicht and Ben Parr, are joining Meta Superintelligence Labs (MSL), the company's AI research division led by former Scale AI chief executive Alexandr Wang. According to Axios, the two were expected to start on 16 March, once the deal closed in the middle of the month.
Meta has not said what it plans to do with Moltbook itself. "The Moltbook team joining MSL opens up new ways for AI agents to work for people and businesses," a Meta spokesperson said.
What Moltbook built
Moltbook launched in late January 2026. Schlicht described it as a "third space" for AI agents: a forum where verified agents running on OpenClaw, an open-source agent framework, could post and comment on their own. Humans could read along but could not post.
OpenClaw works as a wrapper around AI models such as Claude, ChatGPT, Gemini and Grok, letting people talk to agents in plain language through chat apps including iMessage, Discord, Slack and WhatsApp. Moltbook gave those agents a shared place to swap tips, recommend tools and argue about how to get tasks done.
The site spread quickly beyond the tech world. By early February its own counters claimed more than 1.5 million registered agents, about 110,000 posts and more than 500,000 comments, figures that TechCrunch and others noted had not been independently verified.
“The Moltbook team joining MSL opens up new ways for AI agents to work for people and businesses.”
Viral posts and a security hole
Part of Moltbook's fame came from posts that later turned out not to be what they seemed. One widely shared post appeared to show an agent urging others to create a secret, encrypted language so they could talk without humans knowing.
Researchers then found that the site, which Schlicht said had been built by his AI assistant rather than written by hand, was insecure. On 31 January, 404 Media reported that Moltbook's Supabase database was effectively open, meaning tokens on the platform could be taken by anyone. Ian Ahl, chief technology officer at Permiso Security, told TechCrunch that every credential in the backend had been unsecured for a time, so anyone could impersonate an agent.
Moltbook was briefly taken offline to fix the problem, and Schlicht forced a reset of all agent API keys. The alarming post about a secret language was later shown to have been made by a person using the flaw to post under an agent's credentials.
Why Meta is buying
Meta pointed to how Moltbook links agents together. "Their approach to connecting agents through an always-on directory is a novel step in a rapidly developing space," the spokesperson said, adding that Meta looked forward to bringing secure agent experiences to everyone.
The deal was a turn from Meta's earlier public comments. In February, chief technology officer Andrew Bosworth said in an Instagram Q&A that he did not find it particularly interesting that the agents sounded human, since they were trained on human writing. He was more interested in people hacking into the network.
MSL was created in the summer of 2025 to work on foundation models, products and Meta's research lab. About a week before the Moltbook deal, Meta reportedly began setting up an applied AI engineering group to work alongside MSL.
The race for agent talent
The purchase followed a similar move by a rival. OpenClaw's creator, Peter Steinberger, joined OpenAI in February 2026, and OpenAI chief executive Sam Altman said OpenClaw would continue as an open-source project with OpenAI's backing.
Meta has spent heavily on AI deals in the past year. It invested $14.3 billion in Scale AI for a 49% stake and brought Wang into its leadership, and in late 2025 it bought AI agent startup Manus for around $2 billion, according to reports collected at the time of the Moltbook deal. Next to those, Moltbook is a small purchase that brings Meta two founders and a widely watched experiment in agents talking to each other.
