◉ PSYCHOHISTORY

Meta Agentic AI / Superintelligence Labs

mechanismAI & Compute · Media & Managed Opposition
Meta is building an artificial society of 1.5 million AI agents that can act on people's behalf online.
Who they are

Meta's agentic-AI effort, run through its Superintelligence Labs under Alexandr Wang.

What they do

It is Meta assembling a network of autonomous AI agents alongside its control of human social networks.

How it works

Its Moltbook system (March 2026) has 1.5M AI agents but only 17,000 human owners; it also bought Manus AI ($2B), a big stake in Scale AI ($14.3B), and Dreamer, and holds a patent for AI that mimics human social-media behavior to keep inactive users' accounts looking active. This gives Meta the social graph of 3.5B humans plus that of the autonomous agents.

Why it matters

It means one company could control both the human and the artificial online population — and monopolize the behavioral data used to model and predict human society.

The engine's record — word for word
Moltbook (Mar 10 2026): 1.5M AI agents, only 17K human owners, 1.5M API keys exposed. Integrated into MSL under Alexandr Wang. Manus AI ($2B, Dec 2025): general-purpose autonomous agent execution. Scale AI ($14.3B, 49% stake, Jun 2025): data labeling + RLHF. Dreamer (acqui-hire): personalized AI agents. Patent US 12513102B2: simulating human social media behavior — digital ghosts maintaining engagement for inactive users. Meta is constructing a synthetic society capable of acting on behalf of the human population. Controls the social graph of BOTH humans (3.5B via Facebook/Instagram/WhatsApp) AND autonomous systems (Moltbook/MSL/Manus). Monopolizes behavioral data for sociophysical modeling.
Follow the trail
Walk this on the live map →
Part of the Psychohistory engine — 2,426 entities, 6,314 documented connections. Open data, built to be proven wrong.