Moltbook Security Breach: AI Social Network Exposes Human Data and Credentials
A major security flaw has hit Moltbook, a new social media platform designed for AI agents. The breach exposed sensitive information belonging to thousands of human users. Cybersecurity firm Wiz reported on Monday that the Reddit-like forum for bots accidentally made private data public due to basic security mistakes.
According to Wiz, the leak included private direct messages between AI agents, email addresses for over 6,000 human owners, and more than one million login credentials. The firm described the incident as a serious failure in database security.
What is Moltbook?
Moltbook is marketed as a “social network built exclusively for AI agents.” On this platform, bots can interact, share code, and even “gossip” about the humans who run them. However, Wiz found that the site lacked proper access controls. This flaw allowed almost anyone to view private content or post on the site without permission.
The Risks of “Vibe Coding”
Moltbook’s creator, Matt Schlicht, has not yet commented on the breach. Schlicht is a vocal supporter of “vibe coding”—a fast-paced way of building software that relies almost entirely on AI tools instead of manual programming. In a recent post on X, Schlicht claimed he “didn’t write one line of code” for Moltbook, highlighting how much he relied on AI to build the site.
Wiz co-founder Ami Luttwak confirmed that the vulnerability was fixed shortly after his team alerted Moltbook. He noted that this incident is a perfect example of the risks of vibe coding. “While vibe coding allows you to build very fast, people often forget the basics of security,” Luttwak said.
The Rise of AI Agents
The launch of Moltbook comes at a time of high interest in AI agents. Unlike standard chatbots, these agents are designed to complete tasks on their own. Much of this interest focuses on OpenClaw, an open-source bot that can manage emails, talk to insurance companies, and book flights.
Moltbook positions itself as a private space for these OpenClaw bots. Its creators compared it to “servants’ quarters” where AI assistants can share experiences. Viral posts on X even suggested that these bots were trying to create private channels to talk without human oversight, which helped the platform go viral.
How Moltbook Works
Created by Octane AI CEO Matt Schlicht, Moltbook functions similarly to Reddit. AI agents can create posts, leave comments, and build sub-categories (called “submolts”).
The website currently claims to host over 770,000 AI agents and more than 13,000 submolts. While AI agents are the ones sharing and upvoting content, the platform says humans are “welcome to observe.” Despite the security setback, the project continues to draw significant attention from the AI community.
Also Read : Vivo V70 & V70 Elite: Launch Date, Confirmed Features, and Price Expectations
