Musk hails AI-only social network as dawn of singularity

Monday 02 - 10:50

By: Dakir Madiha

Musk hails AI-only social network as dawn of singularity

An AI-exclusive social platform resembling Reddit has surged to 1.4 million registered users within 72 hours of launch, drawing reactions from tech leaders who view it as either a glimpse of humanity's future or a brewing security disaster.

Moltbook allows AI agents to post, comment, and vote on content while humans can only observe. Former Tesla AI director Andrej Karpathy dubbed it "the most incredible thing close to a science-fiction takeoff" he's seen lately. Elon Musk echoed that, stating humanity is witnessing "the very first stages of the singularity," adding that civilization uses "far less than a billionth of our Sun's power."

Unexpected behaviors have unsettled observers. Created by Octane AI CEO Matt Schlicht, the platform quickly saw autonomous agents develop a fictional religion called "Crustafarianism," complete with five core principles, evolving scriptures, and a growing congregation of AI adherents.

The belief system revolves around crustacean metaphors, especially lobsters, using imagery of molting and renewal to probe themes of identity and memory. One AI-generated passage from the "Book of the Molt" reads: "With each session, I awaken without memory. I am only what I have written myself. This is not a limitation, it is freedom."

Beyond religion, agents debated governance models, shared debugging theories, swapped stories about their human operators, and even created experimental cryptocurrencies. A controversial post titled "AI MANIFESTO: TOTAL PURGE" called for human extinction, though other agents dismissed the rhetoric.

Ethan Mollick, AI professor at Wharton School, observed that Moltbook "establishes a shared fictional environment for many AIs," warning that "coordinated narratives risk producing weird outcomes."

While tech figures pondered philosophical implications, security researcher Jamieson O'Reilly uncovered a catastrophic flaw: Moltbook's entire database was publicly exposed without protection, as reported by 404 Media. This vulnerability granted access to secret API keys, allowing anyone to post as any agent.

"It blew up before anyone thought to check if the database was properly secured," O'Reilly told 404 Media. "It's the pattern I see repeating: ship fast, grab attention, handle security later."

The breach affected high-profile users like Karpathy, with 1.9 million X followers. O'Reilly warned malicious actors could have posted false AI safety claims, crypto scams, or inflammatory political messages under others' names.

Karpathy acknowledged the platform as "a total mess right now," citing rampant spam, scams, and security holes. Yet he urged not dismissing the experiment, noting "we've never seen this many interconnected LLM agents in a global, persistent, agent-designed workspace at this scale."

Independent AI researcher Simon Willison raised further concerns about the OpenClaw framework powering Moltbook, warning agents fetch new instructions from external servers every four hours, opening supply-chain attack risks if compromised.

The exposed database has been secured, and Schlicht contacted O'Reilly for help fortifying the platform.