The dark side of AI: ChatGPT's dreaded twin

WormGPT exposes how Al's powerful capabilities are being weaponised, making cybercrime more scalable, persuasive, and dangerously accessible.

- BY GOVIND RAMMURTHY

Every breakthrough brings a shadow.

The printing press gave us literature and propaganda. The internet ushered in connectivity and cybercrime. Now, artificial intelligence has gifted us ChatGPT—and its malicious counterpart, WormGPT.

While millions have marvelled at ChatGPT’s ability to write poetry and solve problems, cybercriminals were developing their own Al assistant. WormGPT is a chilling example of what happens when artificial intelligence meets criminal intent. Promoted on Telegram channels and dark web forums as a “jailbreak-free, unfiltered” alternative to mainstream Al, WormGPT has become the tool of choice for Al-powered cybercrime.

THE CYBERCRIMINAL'S AI ASSISTANT

Unlike ChatGPT, which refuses requests to write malicious code or provide dangerous instructions, WormGPT has no such constraints. For a monthly fee of USD 60-120—roughly the price of a gym membership—cybercriminals can access an Al model that offers capabilities including the creation of advanced malware such as keyloggers, ransomware, and Trojans; the generation of compelling phishing emails designed to deceive even cautious users; and the development of system exploits and reverse shells for infiltration.

It can also analyse code in real time to spot vulnerabilities, scrape websites to extract data or deface pages, and help attackers bypass antivirus detection using advanced obfuscation techniques. In effect, it is a tireless criminal accomplice—available 24/7, requiring no profit share, and accepting cryptocurrency payments.

BUILT FOR DISRUPTION