Методи виявлення та протидії атакам prompt injection у системах на основі великих мовних моделей

Abstract

This work analyzes the problem of prompt injection attacks in LLM-based systems and reviews existing approaches to their detection and mitigation. The research highlights the importance of developing reliable security mechanisms for modern AI-driven systems. Large language models (LLMs) have become one of the most influential technologies in modern artificial intelligence and data mining. They are widely used in intelligent assistants, chatbots, document analysis systems and decision support applications. However, the rapid adoption of LLM-based systems introduces new security challenges. One of the most critical threats is prompt injection, where maliciously crafted prompts manipulate the behavior of a language model and bypass system restrictions.