#delimiter-based-defense

Random Delimiter 도입을 통한 LLM Prompt Injection 방어율 29%p 상승

I Tested Delimiter-Based Prompt Injection Defense Across 13 LLMs