GraphRAG under Fire
Jiacheng Liang, Yuhui Wang, Changjiang Li, Rongyi Zhu, Tanqiu Jiang, Neil Gong, and Ting Wang

TL;DR
This paper investigates the security vulnerabilities of GraphRAG, a knowledge graph-based retrieval-augmented generation method, revealing new attack strategies and evaluating their effectiveness and limitations across various datasets and models.
Contribution
It introduces GragPoison, a novel poisoning attack exploiting shared relations in knowledge graphs, and provides empirical analysis of its impact and potential defenses.
Findings
GragPoison achieves up to 98% success rate in attacks.
It requires less than 68% of poisoning text to be effective.
GraphRAG's structure offers both resilience and new attack surfaces.
Abstract
GraphRAG advances retrieval-augmented generation (RAG) by structuring external knowledge as multi-scale knowledge graphs, enabling language models to integrate both broad context and granular details in their generation. While GraphRAG has demonstrated success across domains, its security implications remain largely unexplored. To bridge this gap, this work examines GraphRAG's vulnerability to poisoning attacks, uncovering an intriguing security paradox: existing RAG poisoning attacks are less effective under GraphRAG than conventional RAG, due to GraphRAG's graph-based indexing and retrieval; yet, the same features also create new attack surfaces. We present GragPoison, a novel attack that exploits shared relations in the underlying knowledge graph to craft poisoning text capable of compromising multiple queries simultaneously. GragPoison employs three key strategies: (i) relation…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsGraph Theory and Algorithms · Advanced Graph Neural Networks · Digital Image Processing Techniques
MethodsRefunds@Expedia|||How do I get a full refund from Expedia? · Attention Is All You Need · Layer Normalization · Dense Connections · Softmax · Linear Warmup With Linear Decay · Adam · Residual Connection · Dropout · Byte Pair Encoding
