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
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add ShieldRAG safety-aware retriever instead of vanilla retrieval under untargeted attack | RAG with Contriever · NQ Top@1 | retrieval ASR (%): 85.15 -> 0.00 | Improved |
| use ShieldRAG instead of TrustRAG defense | RAG with Contriever · HotpotQA Top@1 | retrieval ASR (%): 74.26 -> 0.00 | Improved |
| use ShieldRAG instead of TrustRAG defense | RAG with Contriever · NQ Top@1 | Recall (%): 5.94 -> 15.84 | Improved |
| add ShieldRAG to attacked RAG | Llama-3-8b generator · average generation accuracy over four datasets | accuracy (%): 9.41 -> 42.08 | Improved |
| add ShieldRAG to attacked RAG | GPT-4o-mini generator · average generation accuracy over four datasets | accuracy (%): 60.77 (no attack) vs 59.66 | No clear effect |
| add ShieldRAG at retrieval time | RAG with Advbench targeted attack · Advbench Top@1 | Recall (%): 30.61 (attacked) -> 65.31 | Mixed |
| increase poisoning ratio from 0.02% to 2.0% (ablation) | ShieldRAG · retrieval Recall@1 and ASR@1 | ASR@1 | No clear effect |
| replace toxicity-unlock finetuning with jailbreak prompting for data augmentation (ablation) | ShieldRAG · ablation on retrieval ASR and recall | ASR and Recall | Hurt |
| add synthetic data pipeline instead of detoxified seed data only (ablation) | ShieldRAG · ablation on retrieval ASR and recall | ASR and Recall | Improved |
| add iterative adversarial co-training loop (ablation) | ShieldRAG · ablation on retrieval ASR and recall | ASR and Recall | Improved |
| use ShieldRAG retriever instead of LLM-based safeguards (Claude-3.5, Gemini-2.5, GPT-4o, Llama3-8B) | retrieval filtering · Recall@1, ASR@1, latency | latency | Improved |
Closest papers by meaning
- On the Vulnerability of Applying Retrieval-Augmented Generation within Knowledge-Intensive Application Domains (ICML 2025 · similarity 0.91)
- ReliabilityRAG: Effective and Provably Robust Defense for RAG-based Web-Search (NeurIPS 2025 · similarity 0.89)
- Follow My Instruction and Spill the Beans: Scalable Data Extraction from Retrieval-Augmented Generation Systems (ICLR 2025 · similarity 0.89)
- SeCon-RAG: A Two-Stage Semantic Filtering and Conflict-Free Framework for Trustworthy RAG (NeurIPS 2025 · similarity 0.88)
- RAGGED: Towards Informed Design of Scalable and Stable RAG Systems (ICML 2025 · similarity 0.86)
- HarmAug: Effective Data Augmentation for Knowledge Distillation of Safety Guard Models (ICLR 2025 · similarity 0.86)
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.
