Entity · paper

MIRAGE: Defending Long-Form RAG Against Misinformation Pollution

paperactiveprovisionalmirage-defending-long-form-rag-against-misinformation-pollution-f31126eb·1 events·first seen 38h ago

Aliases: MIRAGE: Defending Long-Form RAG Against Misinformation Pollution

Co-occurring entities

MIRAGE

More like this (12)

MIRAGE Multi-Agentic System Leveraging Open-Source LLMs to Mitigate Disinformation Threats Designed by Journalists, but Is It for Readers? Rethinking AI Disclosures and Transparency in News LightRAG One Polluted Page Is Enough: Evaluating Web Content Pollution in Generative Recommenders ComoRAG Uncertainty-Aware Hybrid Retrieval for Long-Document RAG Privacy-Preserving RAG via Multi-Agent Semantic Rewriting Graph RAG Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning RAG disinformation

Recent events (1)

5arXiv · cs.CL·38h ago·source ↗

MIRAGE: Training-free defense for RAG systems against misinformation in retrieved documents

Researchers introduce MIRAGE, a training-free, model-agnostic defense mechanism for long-form Retrieval-Augmented Generation systems exposed to polluted retrieval corpora. The system builds an NLI-based cross-document claim graph and uses a Defended-Claims Gate to either condition generation on consistently supported evidence or fall back to parametric answering. The authors also release a pollution benchmark protocol spanning four perturbation families and demonstrate that MIRAGE restores factuality across four long-form QA benchmarks where vanilla RAG degrades significantly. Code and datasets are publicly released.

Evaluation and Benchmarking Agent and Tool Ecosystem MIRAGE: Defending Long-Form RAG Against Misinformation Pollution MIRAGE