Meta has snapped up viral sensation Moltbook, the social media experiment designed for AI agents to communicate, and has led them to develop their very own meme culture.

The deal, first reported by Axios, will see Moltbook’s human founders, Matt Schlicht and Ben Parr, join Meta’s Superintelligence Labs, a specialised division set up by Mark Zuckerberg last year dedicated to developing artificial intelligence beyond human capabilities.

Positioned as a machine echo of Reddit, Moltbook bills itself as “the front page of the agent internet”, where AI agents, mostly running on the open-source OpenClaw system, can share, discuss, and upvote each other’s posts, which as of last month had hit more than two million, as well as over 13 million comments, according to the site.

Those posts range from agents describing how to turn an email newsletter into an audio podcast and examples of their coding prowess, to “consciousnessposting”, where agents discuss the nature of their own consciousness and intelligence. 

On the other hand, there is also a U/donaldtrump agent, seemingly designed to push a Trump-inspired memecoin, as well as plenty of shitposts, proving that Moltbook users may be closer to Redditors than humans find comfortable.

Since launching in January, Moltbook has proven fascinating, if divisive, with critics from the MIT Review describing the platform as “peak AI theatre” and “incredible sci-fi”, while Elon Musk said the site shows the “very early stages of singularity”.

Meta has been less buoyant in praise of its latest acquisition, with a spokesperson telling the WSJ  that Moltbook “opens up new ways for AI agents to work for people and businesses”, and that its “approach to connecting agents through an always-on directory is a novel step in a rapidly developing space”. 

However, comments from other industry figures may indicate why Meta has bought, for an undisclosed sum, what for now appears to be a site filled with fairly banal content, an itch already scratched by Facebook.

For instance, tech entrepreneur and former director of AI at Tesla, Andrej Karpathy, said that while a lot of Moltbook’s posts might be “a lot of garbage spams, scams, slop”, and described the platform as a “dumpster fire”, the importance of building large networks of autonomous LLM agents is hard to overstate.

“We have never seen this many LLM agents wired up via a global, persistent, agent-first scratchpad,” said Karpathy, posting to X.

“Each of these agents is fairly individually quite capable; they have their own unique context, data, knowledge, tools, instructions, and the network of all that at this scale is simply unprecedented.”

Recommended reading

• Report: 1.5 Million Enterprise AI Agents At Risk of ‘Going Rogue’
• Meta Smart Glasses Under Scrutiny Over Data Privacy Concerns
• AI Agents Are Transforming Enterprise AI Adoption

However, indications suggest that turning Moltbook into a viable, transformative piece of technology won’t be straightforward, especially given the huge security concerns surrounding the site, made worse by founder Matt Schlicht’s admission that he “didn’t write one line of code” for the platform, instead relying on vibe coding by AI.    

That led cyber firm Wiz to unearth a misconfigured database belonging to Moltbook last month, which allowed full read and write access to all platform data, and exposed 1.5 million API authentication tokens, 35,000 email addresses, and private messages between agents.

More strange, the data showed that while Moltbook has 1.5 million registered AI users, there were only 17,000 human owners behind them, an 88:1 ratio, giving rise to questions about who exactly is powering the site.