A hypothetical scenario could contain an AI-driven customer care chatbot manipulated by way of a prompt containing destructive code. This code could grant unauthorized usage of the server on which the cha