Underground AI models promise to be hackers ‘cyber pentesting waifu’ 

Palo Alto Networks’ Unit 42 reports that underground forums are commercializing custom and jailbroken LLMs—most notably WormGPT 4 and the open-source KawaiiGPT—to aid lower-level hacking tasks such as vulnerability scanning, encryption, data exfiltration, and code generation. WormGPT 4 is sold via inexpensive subscriptions with optional source-code purchase, while KawaiiGPT is freely available on GitHub and maintained by a 500-strong community, illustrating accessible dual-use AI that lowers the technical barrier for attacks even as much of the generated code remains detectable.