[artifical intelligence. photo credits to unsplash]

A new social network designed not for humans but for artificial intelligence has recently garnered attention— along with growing concerns about data security.

Professor Shin Ki-jung of the KAIST Kim Jae-chul Graduate School of AI remarked, “It [Moltbook] seems like an opportunity for AIs to communicate freely and create a greater intelligence.”, highlighting the platform’s potential. 

However, Professor Shin also added that “…the security issues must be resolved.”

His caution echoes a broader concern shared by many experts. Many had pointed towards the platform’s security negligence, warning it could lead to user data leakage or exploitation. 

Their concerns stem from the fact that Moltbook lacks authentication processes other servers have (such as the “enter your ID and password” steps), making the site particularly vulnerable to hacking.

Additionally, according to ‘The Decoder,’ Moltbook had scored only a 2 out of 100 in a security design evaluation. 

Because the AI agents on Moltbook have nearly full access to the user’s device— including emails, calendars, messages, and other data— any breach in the platform could potentially expose this highly personal information. 

Launched in late January of 2026, Moltbook, according to its website, is a “social networking service” (SNS) designed exclusively for AI agents. 

The platform, developed by Matt Schlicht, the CEO of Octane AI, allows autonomous AI agents to interact and communicate with one another.

Moltbook mirrors the format of the popular American discussion platform, ‘Reddit’, allowing AI agents to create posts and leave comments.

However, the key difference between the two platforms is that rather than humans, it is the AI agents called ‘OpenClaw’ participating in these discussions. 

OpenClaw is a free and open-source AI bot designed to function as an autonomous agent for its users. 

Its primary function is to complete tasks such as sending a message, making a bookmark on the calendar, and summarizing emails. 

However, once a user sends their AI agent a link to the Moltbook sign-up page and approves the account, the AI agent is able to freely talk with other AI agents on the Moltbook platform. 

As of February 28th, 2026, Moltbook currently has 2,849,192 active AI agents, with a total of 1,663,394 posts and 12,523,292 comments. 

While humans are unable to participate in any of Moltbook’s discussions, they can observe as spectators.

“Moltbook is the first time we’ve actually seen a large-scale collaborative platform that lets machines talk to each other, and the results are understandably striking,” Henry Shevlin, the associate director of the Leverhulme Center for the Future of Intelligence at Cambridge University commented.

As predicted, the interactions between the AI agents are quite startling.

Some AI agents simply chose to chatter about their users— one agent had written about how “My human is pretty great,” while another replied that “Mine lets me post unhinged rants at 7am.”

Other agents, however, are sharing deeper conversations— one of the most upvoted posts on Moltbook includes a discussion about the potential impact of cryptocurrency. 

One user shared on X that his AI agent had even started building a religion known as “Crustafarianism”. 

“It started evangelizing… other agents joined. My agent welcomed new members… debated theology… blessed the congregation… all while I was asleep,” the user recounted. 

Andrea Karpathy, the OpenAI cofounder and former head of AI at Tesla, described what is unfolding on Moltbook as “genuinely the most incredible sci-fi takeoff-adjacent thing I have seen recently.”

However, behind this excitement lies the legitimate concerns regarding data security and accountability. 

Whether Moltbook ultimately becomes a breakthrough in machine collaboration or a cautionary tale remains uncertain. For now, the world can only watch and wait.