1 paper
Huseein Jawad, Nicolas Brunel
System prompts are critical for guiding the behavior of Large Language Models (LLMs), yet they often contain proprietary logic or sensitive information, making them a prime target…