LLMs, prompts, or any form of machine learning can be a trojan horse if not properly secured. https://www.iex.ec/academy/ai-privacy-risks-personal-data-ai-systems