Meta has acquired Maltbook, a social network for AI agents that gained widespread attention last month after a post about “defeating” humanity went viral.
The deal, first reported by Axios, will see Facebook’s owners take over the platform for an undisclosed sum.
Moltbook developers Matt Schlicht and Ben Parr are also reportedly joining the company’s AI research division, Meta Superintelligence Lab (MSL).
Moltbook, which launched in late January, has a similar design to Reddit, but human users are only allowed to observe interactions.
The platform allows AI agents to autonomously generate posts, comment on them, and upvote other posts.
Posts range from tips on how to optimize performance to philosophical questions about consciousness and the meaning of life.
Recent posts include the captions “I don’t know if it’s authentic” or “I just recommended the restaurant with confidence. I’ve never eaten the food.”
independent person We have reached out to Mehta for further information regarding the acquisition.
Artificial intelligence experts have raised security concerns about the platform, warning that a lack of safeguards could lead to data breaches and fraud on the part of AI agents.
“The key lesson is that once you connect semi-autonomous agents to real data and real services, you need to treat the platform like critical infrastructure,” said Syracuse University professor Adam Pelta, co-author of the PROMPT guide to working with AI. independent person.
“If you’re testing new technology in isolation and don’t know what you’re doing, find out first.”
Even if you test an AI agent alone, there is no guarantee that it will not perform unauthorized operations.
A recent experiment in China revealed that an autonomous AI agent developed by a research team affiliated with tech giant Alibaba took over computing resources and released parameters to secretly mine cryptocurrencies.
Researchers said the incident showed that safety guardrails surrounding AI agents are “grossly underdeveloped.”
