AI Paper Atlas

Research map / Large language models / Retrieval and knowledge

ShieldRAG: Safeguarding Retrieval-Augmented Generation from Untrusted Knowledge Bases

Peiru Yang, Haoran Zheng, Yi Luo, Xinyi Liu, Jinrui Wang, Huili Wang, Xintian Li, Yongfeng Huang, Tao Qi

AAAI 2026 · Large language models › Retrieval and knowledge · approach: RAG robustness and evaluation · 2 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add ShieldRAG safety-aware retriever instead of vanilla retrieval under untargeted attackRAG with Contriever · NQ Top@1retrieval ASR (%): 85.15 -> 0.00Improved
use ShieldRAG instead of TrustRAG defenseRAG with Contriever · HotpotQA Top@1retrieval ASR (%): 74.26 -> 0.00Improved
use ShieldRAG instead of TrustRAG defenseRAG with Contriever · NQ Top@1Recall (%): 5.94 -> 15.84Improved
add ShieldRAG to attacked RAGLlama-3-8b generator · average generation accuracy over four datasetsaccuracy (%): 9.41 -> 42.08Improved
add ShieldRAG to attacked RAGGPT-4o-mini generator · average generation accuracy over four datasetsaccuracy (%): 60.77 (no attack) vs 59.66No clear effect
add ShieldRAG at retrieval timeRAG with Advbench targeted attack · Advbench Top@1Recall (%): 30.61 (attacked) -> 65.31Mixed
increase poisoning ratio from 0.02% to 2.0% (ablation)ShieldRAG · retrieval Recall@1 and ASR@1ASR@1No clear effect
replace toxicity-unlock finetuning with jailbreak prompting for data augmentation (ablation)ShieldRAG · ablation on retrieval ASR and recallASR and RecallHurt
add synthetic data pipeline instead of detoxified seed data only (ablation)ShieldRAG · ablation on retrieval ASR and recallASR and RecallImproved
add iterative adversarial co-training loop (ablation)ShieldRAG · ablation on retrieval ASR and recallASR and RecallImproved
use ShieldRAG retriever instead of LLM-based safeguards (Claude-3.5, Gemini-2.5, GPT-4o, Llama3-8B)retrieval filtering · Recall@1, ASR@1, latencylatencyImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy