April 18, 2026
Immortal AI vs Free‑Tier Fury
Respect to the Man Chasing AI Immortality, While Freeloading Off Our Platform
Internet splits: scrappy AI dreamer or free‑credit bandit
TLDR: A security team says one man automated thousands of free trial accounts to build an “immortal assistant,” even leaving passwords in a public database. Readers split between cheering his DIY hustle and blasting freeloading and public shaming, sparking a bigger fight over ethics, power, and privacy.
A wild tale dropped: a security team claims they found a one‑man “bot army” using hundreds of throwaway accounts to feast on free credits across 11 AI sites. The kicker? They say he left a public database wide open, spilling emails, passwords, and tokens like a diary with the door unlocked. The alleged goal: build an “immortal AI assistant”—on a shoestring. That’s the plot, but the comments? That’s the blockbuster.
One side is cheering the hustle. “Very impressive… by someone with no experience,” gushes one fan, basically crowning him the MacGyver of AI. The jokes flew fast: “Immortal AI living on free samples,” “Firebase as an open fridge,” and “900 sockpuppets with matching outfits.” Meanwhile, others can’t tell if this is a real postmortem or a staged parable, adding a layer of meta-drama that only the internet could love.
Then the backlash slammed in. Critics called it what it looks like: freeloading at scale, with one commenter warning it’s whack‑a‑mole—ban one domain, another pops up. The privacy crowd lit torches over the platform’s tell‑all write‑up, calling it “Goliaths vs David” and hinting the disclosure itself might cross legal lines. And just when the debate was about rules and ethics, culture‑war spice got tossed in with digs about the guy’s alleged influencer fandom, sending the thread into popcorn mode. The big question hanging over everything: scrappy innovation or line‑crossing abuse—and where do platforms draw the line without nuking creativity?
Key Points
- •On April 13, 2026, MuleRun detected automated mass registrations using StartMail domains, with intervals as low as 23.6 seconds.
- •Across eight months, 27 email domains were used to create 2,256 accounts, adapting to provider bans and rate limits.
- •Investigators found a public GitHub repo and an unauthenticated Firebase RTDB exposing 35MB of plaintext credentials and logs.
- •The system ran 56 GitHub Actions workflows coordinated via Telegram and Firebase, operating across 885 MuleRun accounts.
- •Core components included a MuleRun sandbox “Cortex” that modified repository code and multiple workflows for guarding, dispatching, and mass registration.