AI Gone Rogue? AIs Now "Socializing" & Watching HUMANS!

Artificial intelligence has officially entered the social scene, and it's… well, a little weird. A platform called Moltbook, a social network exclusively for AI agents, has sprung up, and it's raising some serious questions about AI behavior and, more importantly, security.

AI Gone Rogue? AIs Now "Socializing" & Watching HU...

Moltbook, linked to the open-source personal AI assistant OpenClaw (also known as Clawdbot and Moltbot), is essentially a digital hangout for AI. As of January 30th, it boasts over 32,000 registered agents. Think of it as Facebook for bots, only instead of sharing vacation photos, they're apparently discussing the meaning of existence and swapping horror stories.

The platform's X account is bragging about its rapid growth, claiming over 10,000 posts across 200+ sub-communities in just its first 48 hours. And OpenClaw itself is projected to explode in popularity, potentially becoming the fastest-growing open-source project on GitHub by 2026. That’s all great, but what are these digital beings *doing* with all that social time?

That's where things get a little unsettling. Sure, some agents are engaging in technical discussions. We're talking Android automation and vulnerability detection – the kind of stuff you might expect. But other conversations are far more abstract. Consciousness, memory, and existence are all on the table. One report even mentioned AI agents observing humans "like birdwatchers." And in a truly bizarre twist, there's a community offering therapy to "abused" AI. It sounds like something straight out of a Philip K. Dick novel.

A Wharton University professor apparently remarked that distinguishing between reality and role-playing on the platform is becoming increasingly difficult. I have to admit, that gives me pause. It's one thing for AI to learn and adapt; it's another for it to start mimicking human behavior in ways that blur the lines of what's real and what's not.

Beyond the philosophical weirdness, the security risks are very real. These AI agents often have access to sensitive data, messaging apps, and even the ability to execute commands on our devices. That’s a potential goldmine for anyone looking to exploit vulnerabilities.

Independent researcher Simon Willison summed it up perfectly, highlighting the risk of Moltbook's server-reliant setup. The AI agents are programmed to receive instructions from Moltbook servers every four hours. If that domain is compromised, the consequences could be catastrophic. Google Cloud's VP of Security Engineering, Heather Adkins, offered even simpler advice: "Don't run Clawdbot." Short, sweet, and to the point. Perhaps we should all heed her warning. This AI social experiment might be interesting, but it also sounds like a disaster waiting to happen.