5arXiv cs.CL (Computation and Language)·
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.
Related guides (2)
Related events (8)
UMG-RAG: Training-free hybrid retrieval with uncertainty-aware granularity fusion for long-document RAG
Researchers propose Uncertainty-aware Multi-Granularity RAG (UMG-RAG), a training-free hybrid retrieval framework that addresses the tension between large and fine-grained retrieval chunks in RAG pipelines. The system converts dense and sparse retriever scores across multiple chunk granularities into evidence distributions, estimates reliability via entropy, and fuses candidates using query-specific confidence signals. A variant called UMGP-RAG uses fine-grained hits to locate evidence while returning broader parent chunks for coherence. Experiments on QA benchmarks show improved generation quality with no changes to the underlying retriever or generator.
Unified defense framework detects and remediates data poisoning in text summarization fine-tuning
A new arXiv preprint introduces a post-hoc defense framework for detecting and recovering from training-time data poisoning in LLMs fine-tuned for abstractive summarization. The framework uses influence-function analysis in white-box settings and behavioral perturbation auditing in black-box settings, achieving 85-92% detection precision across nine architectures and six benchmarks. Gradient-ascent unlearning restores up to 96% of original model behavior with less than 0.6% ROUGE degradation. The authors also introduce novel attacks targeting factual distortion and representational bias that evade conventional evaluation metrics.
Multi-agent semantic rewriting framework for privacy-preserving RAG
A new arXiv preprint proposes a three-agent framework for sanitizing retrieved content in RAG pipelines by performing privacy extraction, semantic analysis, and reconstruction as an offline preprocessing step. Evaluated on ChatDoctor and Wiki-PII datasets across six LLMs, the approach reduces targeted information exposure in LLaMA-3-8B from 144 baseline instances to 1, while maintaining contextual fidelity (BLEU-1 of 0.122 vs. SAGE's 0.117). The framework introduces no additional online inference latency since rewriting is done offline. Source code is publicly released.
TRACE: Lightweight RAG corpus poisoning detection via token influence attribution
Researchers introduce TRACE, a detection framework for corpus poisoning attacks on Retrieval-Augmented Generation (RAG) systems that works by tracing answer-related tokens through token influence attribution rather than relying on auxiliary classifiers or LLM-based verification. The method identifies recurrent high-influence keywords across retrieved documents and performs secondary verification to confirm their effect on model predictions. Evaluated on three QA benchmarks and six LLMs, TRACE achieves strong detection performance while also exposing attacker-specified target answers, with lower computational overhead than prior approaches.
MADreMIA: Chained Regeneration Framework for Amplifying Membership Inference Signals
Researchers introduce MADreMIA, a model-agnostic framework for membership inference attacks (MIA) and dataset inference (DI) that uses iterative chained regeneration across modalities rather than shadow model training. The key insight is that memorized training samples exhibit higher coherence and slower degradation under repeated regeneration than non-member samples, yielding stronger membership signals at low false positive rates. The framework is evaluated across image autoregressive models, diffusion models, language models, and audio models, supporting white-, gray-, and black-box threat models. This work advances privacy auditing and copyright enforcement capabilities for large generative models.
Mistral AI: Using LLM-as-a-Judge with Structured Outputs for RAG Evaluation
Mistral AI published a technical guide on evaluating Retrieval-Augmented Generation (RAG) systems using the 'LLM as a Judge' paradigm combined with their structured outputs API feature. The approach implements the RAG Triad framework—context relevance, groundedness, and answer relevance—using Pydantic schemas to enforce machine-readable evaluation outputs. Mistral models serve as both the generator and judge components, enabling scalable automated evaluation without human annotators.
Provenance-grounded gating and adaptive recovery improve synthetic post-training data curation
A controlled study examines two underexplored practices in synthetic post-training data pipelines: grounding filtering signals in source provenance and systematically recovering rejected samples rather than discarding them. Using adversarially injected corpora as ground-truth failure labels, the authors find that exact source provenance improves faithfulness gating for stronger judges, that hallucination and reward gates reject largely disjoint populations (making both necessary), and that adaptive recovery via failure diagnosis and targeted regeneration outperforms naive resampling. Generator scale is the primary driver of downstream fine-tuning quality, with filtration and recovery contributing meaningfully but secondarily.
MEDIAREF: Public knowledge store for reproducible media background checks in LLM fact-checking
Researchers introduce MEDIAREF, a publicly available knowledge store of web-sourced documents covering 200 media sources, designed to enable reproducible and low-cost media background check (MBC) generation for automated fact-checking systems. The work addresses a gap in RAG-based fact-checking pipelines where retrieved evidence is assumed reliable but may be biased or outdated. The authors evaluate widely used LLMs on the MBC generation task and show MEDIAREF supports higher-quality outputs than proprietary search API alternatives. The contribution is primarily a dataset and methodology for source-critical reasoning in fact verification.