Cybercriminals no longer play cat and mouse — they just create their own LLMs without filters that do any dirty work.

A whole class of 'anti-GPT' models has surfaced on black markets, and researchers have already studied two such combat neural networks.

💀 WormGPT 4

Sold on Telegram and dark web platforms by subscription for $50–$220.

This machine does things that normal models won't even attempt:

— generates phishing emails at the level of premium banks,

— writes malware in Python,

— helps to assemble a full-fledged malware even for a person who didn’t know where the Enter key was yesterday.

☠ KawaiiGPT

And this is where the real trash begins:

the model is free, available on GitHub, and can be set up in 5 minutes.

A simple CLI turns a schoolboy into 'cyber special forces':

— generates phishing bait,

— creates compromise schemes via SSH,

— collects a ransomware chain: encryption → instructions for the victim → payment in crypto → done.

⚙ No ethics. No limits. No blocks.

These models are specifically designed for attacks: phishing, hacks, injections, malware — a complete service packaged in LLM.

🧬 The result is unpleasant but honest:

a new level of threats is already here.

Cybercrime now requires no skills.

It's enough to download the neural network, enter the correct request — and any schoolboy turns into a 'hacker'.

#ïžâƒŁ #CyberSecurity #AÄ° #malware #darkweb

😉 Subscribe to not miss the most dangerous trends in technology.