WormGPT: A Large Language Model Chatbot for Criminals
Mohamed Firdhous, Walid Elbreiki, Ibrahim Abdullahi, B. H. Sudantha, Rahmat Budiarto
Abstract
In recent years, AI has made many advances in multiple fronts. One such recent advance is the development of Large Language Models(LLMs), a deep learning model trained with a large set of data on billions of parameters. Many Generative Pretrained Transformer (GPT) models have been developed on these LLMs. ChatGPT has become the most popular GPT used by many people. Recently, a blackhat GPT named WormGPT has been launched for cybercriminals. WormGPT is capable of launching many types of attacks resulting in irreparable damage. Hence, Users are required to be familiar with these tools, if they are to protect themselves. This article is possibly the very first attempt to initiate a discussion on this topic within the formal research circles. This article starts with an introduction to LLMs and GPTs and then presents the details of WormGPT, its capabilities and the damage it can cause. Finally, a set of safeguards has been presented that can help to protect users from possible attacks.