CLAug 5, 2025

Token-Level Precise Attack on RAG: Searching for the Best Alternatives to Mislead Generation

arXiv:2508.03110v1h-index: 9
Originality Highly original
AI Analysis

This addresses a critical security problem for RAG systems, which are used to enhance LLMs with external knowledge, by revealing vulnerabilities that could be exploited to manipulate outputs.

The paper tackles the security vulnerability in retrieval-augmented generation (RAG) systems by proposing TPARAG, a token-level attack framework that outperforms previous methods in retrieval and end-to-end attack effectiveness on open-domain QA datasets.

While large language models (LLMs) have achieved remarkable success in providing trustworthy responses for knowledge-intensive tasks, they still face critical limitations such as hallucinations and outdated knowledge. To address these issues, the retrieval-augmented generation (RAG) framework enhances LLMs with access to external knowledge via a retriever, enabling more accurate and real-time outputs about the latest events. However, this integration brings new security vulnerabilities: the risk that malicious content in the external database can be retrieved and used to manipulate model outputs. Although prior work has explored attacks on RAG systems, existing approaches either rely heavily on access to the retriever or fail to jointly consider both retrieval and generation stages, limiting their effectiveness, particularly in black-box scenarios. To overcome these limitations, we propose Token-level Precise Attack on the RAG (TPARAG), a novel framework that targets both white-box and black-box RAG systems. TPARAG leverages a lightweight white-box LLM as an attacker to generate and iteratively optimize malicious passages at the token level, ensuring both retrievability and high attack success in generation. Extensive experiments on open-domain QA datasets demonstrate that TPARAG consistently outperforms previous approaches in retrieval-stage and end-to-end attack effectiveness. These results further reveal critical vulnerabilities in RAG pipelines and offer new insights into improving their robustness.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes