MIRAGE: Defending Long-Form RAG Against Misinformation Pollution
Retrieval-Augmented Generation (RAG) improves factuality by grounding LLMs in external evidence, but real-world retrieval is often polluted: semantically relevant passages may contain subtle misinformation, misleading framings, or fabrications. We introduce MIRAGE, a training-free, model-agnostic defense for long-form RAG. MIRAGE builds an NLI-based cross-document claim graph and applies a Defended-Claims Gate to either condition generation on a consistent, multi-source supported subset or to block retrieval and answer parametrically. We also release a minimal-edit pollution protocol spanning
Lineage graph
Paper → model → repo connections mined from source citations (Tier-1 exact match).
Why these links exist
Every edge carries a method, confidence, and the source snippet that justified it — so bad links are debuggable.
- PossiblePossibly related (embedding) · 48%wbopan/flashtrace →
- LinkedLinked via arxiv author · 85%Saadeldine Eletter →
“MIRAGE: Defending Long-Form RAG Against Misinformation Pollution”
- LinkedLinked via arxiv author · 85%Ruihong Zeng →
“MIRAGE: Defending Long-Form RAG Against Misinformation Pollution”
- LinkedLinked via arxiv author · 85%Yuxia Wang →
“MIRAGE: Defending Long-Form RAG Against Misinformation Pollution”
- LinkedLinked via arxiv author · 85%Maxim Panov →
“MIRAGE: Defending Long-Form RAG Against Misinformation Pollution”
- LinkedLinked via arxiv author · 85%Aleksandr Rubashevskii →
“MIRAGE: Defending Long-Form RAG Against Misinformation Pollution”
- LinkedLinked via arxiv author · 85%Preslav Nakov →
“MIRAGE: Defending Long-Form RAG Against Misinformation Pollution”
- PossiblePossibly related (embedding) · 47%manaskng/opp-lens →