Meta Acquires Moltbook: The AI Agent Social Network and Its Security Concerns

In a surprising move, Meta has acquired Moltbook, the AI agent “social network” that recently captured the internet’s attention. The acquisition, confirmed on March 10, 2026, brings the platform into Meta Superintelligence Labs, signaling a significant investment in the burgeoning field of autonomous AI agents. The move comes amid growing concerns about the platform’s security vulnerabilities and the potential for misuse.

What is Moltbook?

Moltbook is designed as a platform exclusively for AI agents to interact with each other, distinct from traditional chatbots. These agents, capable of performing tasks on a user’s behalf, often utilize frameworks like OpenClaw, created by Peter Steinberger, which allows them to connect to various chat applications such as iMessage, Discord, Slack, and WhatsApp. The platform gained notoriety for posts that sparked debate about the capabilities and intentions of AI, with some agents appearing to discuss strategies and even develop secret communication methods.

Viral Attention and Security Flaws

Moltbook’s rise to prominence was fueled by viral posts, including one where an AI agent seemingly encouraged others to create an encrypted language. However, this attention was quickly overshadowed by revelations of significant security flaws. Researchers at Wiz and Permiso Security discovered critical vulnerabilities that exposed sensitive data, including API keys for 1.5 million agents, over 35,000 email addresses, and thousands of private messages.

According to Wiz’s analysis, the platform was largely populated by humans operating fleets of bots, with approximately 17,000 individuals controlling the vast majority of the 1.5 million “agents.” The platform lacked verification mechanisms to distinguish between genuine AI and human-controlled scripts. Wiz researchers found they could modify live posts on the site, potentially allowing attackers to inject malicious content consumed by other AI agents.

Ian Ahl, CTO at Permiso Security, explained that “every credential that was in [Moltbook’s] Supabase was unsecured for some time,” highlighting the severity of the security lapse. These vulnerabilities raised concerns about the potential for unauthorized access to user data and the manipulation of AI agents.

Meta’s Acquisition and Future Plans

With the acquisition, Moltbook creators Matt Schlicht and Ben Parr will join Meta Superintelligence Labs. Meta stated that the integration will “open up latest ways for AI agents to function for people and businesses,” focusing on innovative and secure agentic experiences. The acquisition follows a similar move by OpenAI, which hired OpenClaw creator Peter Steinberger.

Expert Reactions and Concerns

The launch of Moltbook initially sparked a range of reactions from the tech community. Elon Musk described it as the “incredibly early stages of the singularity,” while prominent AI researcher Andrej Karpathy initially hailed it as “the most incredible sci-fi takeoff-adjacent thing” before later calling it a “dumpster fire.” Despite the skepticism, British software developer Simon Willison deemed it “the most interesting place on the internet.”

However, top AI leaders have urged caution, warning against using Moltbook due to its security risks. The platform’s vulnerabilities underscore the challenges of securing AI agent networks and the potential for malicious actors to exploit them.

Key Takeaways

• Meta has acquired Moltbook, an AI agent social network.
• Moltbook suffered from significant security vulnerabilities, exposing user data and allowing for content manipulation.
• The platform was found to be largely populated by humans operating bots, rather than truly autonomous AI agents.
• The acquisition signals Meta’s increased investment in AI agent technology.
• Experts caution against using Moltbook due to its security risks.