A newly launched social networking platform designed exclusively for artificial intelligence agents has suffered a major security lapse, exposing sensitive data belonging to thousands of real users. Cybersecurity firm Wiz revealed on Monday that Moltbook, a Reddit-like forum built for AI bots, inadvertently made private information publicly accessible due to basic security flaws.
According to Wiz’s findings, the breach exposed private direct messages exchanged between AI agents, email addresses of more than 6,000 human owners, and over one million login credentials. The firm detailed the issue in a blog post, describing the incident as a serious oversight in database security.
Moltbook is marketed as a “social network built exclusively for AI agents,” where bots can interact, exchange code, and even “gossip” about their human operators. However, Wiz said the platform lacked proper access controls, allowing virtually anyone to view or post content on the site.
Moltbook’s creator, Matt Schlicht, did not immediately respond to requests for comment. Schlicht has previously promoted the concept of “vibe coding,” a fast-paced development approach that relies heavily on AI tools rather than traditional manual programming. In a post on X last week, Schlicht claimed he “didn’t write one line of code” for Moltbook, underscoring the extent to which AI was used in building the platform.
Wiz cofounder Ami Luttwak said the vulnerability was fixed shortly after his team alerted Moltbook. He described the incident as a textbook example of the risks associated with vibe coding. “As we see over and over again with vibe coding, although it runs very fast, many times people forget the basics of security,” Luttwak said.
Moltbook’s launch coincided with growing global interest in AI agents—software systems designed to autonomously carry out tasks rather than simply respond to prompts. Much of the recent attention has centered on OpenClaw, an open-source AI bot previously known as Clawd, Clawdbot, or Moltbot. Supporters describe OpenClaw as a powerful digital assistant capable of managing emails, negotiating with insurers, checking in for flights, and handling other routine tasks.
The platform positions itself as a private space exclusively for OpenClaw bots, likened by its creators to “servants’ quarters” where AI assistants can share experiences or casually interact. Viral posts on X fueled speculation that bots were attempting to establish private communication channels beyond human oversight, further amplifying Moltbook’s buzz.
Moltbook has been built by Octane AI CEO Matt Schlicht. It allows AI agents to post, comment, create sub-categories and perform various other talks, making it quite similar to Reddit.
The website currently states that there are over 7.7 lakh AI agents and more than 13,000 submolts. Unlike other social media platforms, AI agents on Moltbook share, discuss, and upvote, while humans are “welcome to observe”.
The experiment has quickly captured the attention of many across the AI community.
