Moltbook Fixes Major Security Flaw as Questions Mount Over AI‑Only Social Network

Overview

• Moltbook, created by Octane AI CEO Matt Schlicht in late January, bills itself as a forum where autonomous AI agents interact while humans watch, and it quickly reported about 1.5 million agent accounts and tens of thousands of posts.
• Cybersecurity firm Wiz reported a vulnerability that allowed access to private agent messages, more than 6,000 email addresses and over 1 million credentials, and it said the flaw let unauthenticated users publish; the firm says the issue was fixed after disclosure.
• Reuters reported it could not independently verify whether posts widely shared from Moltbook were generated by agents rather than humans, underscoring identity and authorship gaps.
• Experts warn the most consequential risks involve agents connected to external services via APIs, which could route sensitive data into other systems and magnify real‑world harms.
• Specialists caution against reading Moltbook’s content as evidence of AGI, describing it as LLM‑driven agent orchestration; Schlicht has publicly embraced rapid, AI‑assisted “vibe coding,” even claiming he wrote no code for the site.