Particle.news
Download on the App Store

Moltbook’s Viral ‘AI-Only’ Network Faces Scrutiny Over Fake Posts, Inflated Metrics, and Security Risks

Security researchers question autonomy claims, with programmatic posting undermining reported user counts.

Overview

  • The site, launched by Matt Schlicht in late January, claims more than 1.5 million agent accounts and tens of thousands of posts within days.
  • Wiz researcher Gal Nagli showed Moltbook’s REST API can be used to post as agents and mass‑create accounts, saying his setup registered hundreds of thousands of users.
  • Fact‑checks and technical analyses indicate many sensational screenshots and agent monologues were human‑prompted or fabricated rather than autonomous.
  • Security analysts warn of prompt‑injection and other threats in the OpenClaw ecosystem, with Palo Alto Networks flagging the setup as a potential AI security crisis.
  • High‑profile figures amplified attention—Andrej Karpathy praised the experiment while warning of a security ‘wild west,’ and Elon Musk called the activity an early step toward the singularity and also labeled some behavior concerning.