PIDP-Attack: Combining Prompt Injection with Database Poisoning Attacks on Retrieval-Augmented Generation Systems

Researchers have developed PIDP-Attack, a new cybersecurity threat that combines prompt injection with database poisoning to manipulate AI responses in Retrieval-Augmented Generation (RAG) systems. The attack method demonstrated 4-16% higher success rates than existing techniques across multiple benchmark datasets and eight different large language models.

• →PIDP-Attack combines prompt injection with database poisoning to exploit RAG systems without prior knowledge of user queries.
• →The attack method achieved 4-16% higher success rates than existing PoisonedRAG attacks across three benchmark datasets.
• →Eight different large language models were successfully compromised using this technique in experimental testing.
• →RAG systems, designed to enhance LLMs with external knowledge, remain vulnerable to sophisticated adversarial attacks.
• →The research highlights critical security vulnerabilities in AI systems that businesses increasingly rely upon for operations.