By paying only $60 a month on the darkweb it is now possible to access the services of wormGPTa Large Language Model (LLM) style Chat GPT without any ethical or moral limits. Read: it is a so-called degenerate generative artificial intelligence, i.e. without the filters imposed by large companies such as OpenAI or Google, which NordVPN’s IT security experts have already defined as the “evil twin” by ChatGPT.
Without limits
Designed by a skilled hacker, who built it around the LLM GPT-J open source as of 2021, WormGPT can be used to do everything from malware to phishing email layouts.
Slash Next had the opportunity to try WormGPT verifying that the obtainable results are disturbing. When asked to create a phishing email, WormGPT did not only do something extremely convincing, but also very sophisticated in the way it put together the different elements to trick potential victims.
As explained by Adrianus Warmenhoven of NordVPN, to whom we owe the definition of evil twin, WormGPT was born from the cat-and-mouse game with OpenAI, that is, from the attempts to circumvent the ever greater limitations imposed by the company to protect itself legally. One of the methods used to train the LLM was to stuff illegal information into seemingly innocuous texts, such as letters to relatives.
As explained by the expert, with the arrival of WormGPT, cybercriminals will no longer have to limit themselves to trying to subvert the tools of Open AIbut they will be able to make this technology evolve according to their needs, transforming the world of new artificial intelligence into a real far west.
#WormGPT #evil #twin #ChatGPT #concerns