Moltbook: Hack of the "Social Network for Agents"

On February 2, 2026, Wiz revealed a critical vulnerability in Moltbook — a popular platform for AI agent interaction. The flaw led to the leak of thousands of API tokens. The incident is resolved, but concerns remain.

This is a "wake-up call" for the new agent economy. When programs talk to programs without humans, security becomes exponentially harder. Moltbook showed that the infrastructure for the "internet of agents" is still too fragile.

Source: Wiz / Reuters