A hypothetical scenario could require an AI-driven customer care chatbot manipulated via a prompt containing malicious code. This code could grant unauthorized usage of the server on which the chatbot ope