Adversarial Threat Vectors and Risk Mitigation for Retrieval-Augmented Generation Systems
This addresses security risks in RAG systems used in industry, but it is incremental as it builds on known threats without introducing new methods.
The paper tackles the vulnerability of Retrieval-Augmented Generation (RAG) systems to adversarial attacks such as prompt injection, data poisoning, and query manipulation, and proposes a prioritized control list including input validation and adversarial training for risk mitigation.
Retrieval-Augmented Generation (RAG) systems, which integrate Large Language Models (LLMs) with external knowledge sources, are vulnerable to a range of adversarial attack vectors. This paper examines the importance of RAG systems through recent industry adoption trends and identifies the prominent attack vectors for RAG: prompt injection, data poisoning, and adversarial query manipulation. We analyze these threats under risk management lens, and propose robust prioritized control list that includes risk-mitigating actions like input validation, adversarial training, and real-time monitoring.