Perhaps because the chatbot is advising the user on how to act in a given situation... and how to circumvent having their chatGPT prompts flagged by the system.