Malicious LLMs empower inexperienced hackers with advanced tools
Unrestricted large language models (LLMs) such as WormGPT 4 and KawaiiGPT are increasing their capacity to produce malicious code, providing useful scripts for lateral movement and ransomware encryptors.
The two LLMs that hackers are increasingly using through paid subscriptions or free local instances were tested by researchers at Palo Alto Networks Unit42.
Although the WormGPT model first appeared in 2023, the project was allegedly abandoned the same year. The brand made a comeback in September with WormGPT 4. It functions as an uncensored ChatGPT variation specifically trained for cybercrime activities and costs $50 per month or $220 for lifetime subscription.
KawaiiGPT is a community-driven, free alternative that was discovered in July of this year. It can create well-written...