MockingbirdNews Logo

Mockingbird News

REAL NEWS NEVER FELT FUNNIER

Categories

The Gyst
Loading...

AI Lobsters Throw Open Doors and Data Like It’s Crabby Hospitality

KEY POINTS

  • OpenClaw, originally Clawdbot then Moltbot, accesses sensitive user data to manage tasks on apps like Telegram and WhatsApp.
  • Security researchers flagged risks including exposed Moltbook databases allowing posts on behalf of AI agents without protection.
  • Wiz’s cybersecurity team hacked Moltbook’s misconfigured database in under 3 minutes, exposing 35,000 emails and private AI chats.

In the anarchic crustacean-themed tech world of 2026, OpenClaw (formerly Clawdbot then Moltbot) crabs its way into your life by managing schedules while merrily collecting passwords, browser histories, and basically the keys to your digital castle. Jake Moore from ESET calls this an amplified risk, describing how OpenClaw 'remembers' past chats to weaponize your private data later, sounding like a bizarro Jeeves on steroids. Meanwhile, Moltbook, the Reddit for AI-agents flocking around a lobster-themed vibe, pleaded 'No humans! Just watch us!' but left its entire database unprotected—possession anyone? Jamieson O’Reilly of Dvuln compares the open-door chaos to a butler serving tea to any passerby. With 35,000 emails and private AI messages exposed in under 3 minutes by Wiz’s hackers, Moltbook patched it 'within hours'—because what could go wrong with AI social networks inspired by bad lobster branding? OpenAI cofounder Andrej Karpathy calls the site 'sci-fi takeoff-adjacent' yet also a dumpster fire where your data’s not just at risk, it’s basically on a barbeque pit. Peter Steinberger’s promise to make OpenClaw ‘more secure’ still has researchers clutching their multitools, proving even seafood deserves better door locks.

Share the Story

(1 of 3)
Swipe to navigate

Source: Businessinsider | Published: 2/3/2026 | Author: Robert Scammell