M³Att poisons medical multimodal RAG by pairing covert textual misinformation with query-agnostic visual perturbations that increase retrieval of the bad content, causing LLMs to generate clinically plausible but incorrect responses.
Towards trustworthy re- trieval augmented generation for large language models: A survey
12 Pith papers cite this work. Polarity classification is still indexing.
12
Pith papers citing it
abstract
Retrieval-Augmented Generation (RAG) is an advanced technique designed to address the challenges of Artificial Intelligence-Generated Content (AIGC). By integrating context retrieval into content generation, RAG provides reliable and up-to-date external knowledge, reduces hallucinations, and ensures relevant context across a wide range of tasks. However, despite RAG's success and potential, recent studies have shown that the RAG paradigm also introduces new risks, including robustness issues, privacy concerns, adversarial attacks, and accountability issues. Addressing these risks is critical for future applications of RAG systems, as they directly impact their trustworthiness. Although various methods have been developed to improve the trustworthiness of RAG methods, there is a lack of a unified perspective and framework for research in this topic. Thus, in this paper, we aim to address this gap by providing a comprehensive roadmap for developing trustworthy RAG systems. We place our discussion around five key perspectives: reliability, privacy, safety, fairness, explainability, and accountability. For each perspective, we present a general framework and taxonomy, offering a structured approach to understanding the current challenges, evaluating existing solutions, and identifying promising future research directions. To encourage broader adoption and innovation, we also highlight the downstream applications where trustworthy RAG systems have a significant impact.
citation-role summary
background 2
citation-polarity summary
roles
background 2polarities
background 2representative citing papers
Augmenting LLM search judges with historical QRI cards improves Spearman correlation with user preferences by ~5% overall (91% relative on disagreements) and 15% in multilingual settings, with better alignment to live A/B test outcomes.
mKG-RAG constructs multimodal KGs via MLLM-driven extraction and vision-text matching then applies dual-stage query-aware retrieval to achieve new state-of-the-art results on knowledge-based VQA.
Retrieval-state lock-in causes zero-dispersion errors in 42% of KG-RAG and 59% of dense-retrieval failures; a three-object check rule reaches 91.9% pooled precision at 7.7% coverage.
BadRDM is a backdoor attack on retrieval-augmented diffusion models that poisons the retrieval database with toxicity surrogates and uses multimodal contrastive learning to force toxic generations from text triggers while preserving benign performance.
Hybrid RAG over UK public health guidance sharply raises MCQA accuracy and free-form faithfulness, letting smaller open models match larger closed models without retrieval.
Proposes an evaluation framework using parametric override and confidence metrics to assess RAG robustness to clean, poisoned, and mixed retrieval evidence on factoid questions.
SLOT organizes RAG security literature by attack Surface, defense Layer, CIA Objective, and Target scope, exposing mismatches between attacks and defenses along a six-stage knowledge pipeline.
MemOS introduces a unified memory management framework for LLMs using MemCubes to handle and evolve different memory types for improved controllability and evolvability.
AstroRAG adds PageRank re-ranking after MMR retrieval inside transient per-instance indexes to improve LLM answers on astronomy questions, reaching 79.49 percent accuracy and F1 on AstroQA with Mistral-7B.
HUMBR reduces LLM hallucinations in enterprise workflows by using a hybrid semantic-lexical utility within minimum Bayes risk decoding to identify consensus outputs, with derived error bounds and reported outperformance over self-consistency on benchmarks and production data.
A survey of architectures, threats, defenses, and future directions for security and privacy in RAG systems.
citing papers explorer
-
Knowledge Poisoning Attacks on Medical Multi-Modal Retrieval-Augmented Generation
M³Att poisons medical multimodal RAG by pairing covert textual misinformation with query-agnostic visual perturbations that increase retrieval of the bad content, causing LLMs to generate clinically plausible but incorrect responses.
-
As It Was: Aligning LLM Search Evaluation with Historical User Preferences
Augmenting LLM search judges with historical QRI cards improves Spearman correlation with user preferences by ~5% overall (91% relative on disagreements) and 15% in multilingual settings, with better alignment to live A/B test outcomes.
-
mKG-RAG: Leveraging Multimodal Knowledge Graphs in Retrieval-Augmented Generation for Knowledge-intensive VQA
mKG-RAG constructs multimodal KGs via MLLM-driven extraction and vision-text matching then applies dual-stage query-aware retrieval to achieve new state-of-the-art results on knowledge-based VQA.
-
When Confidence Takes the Wrong Path: Diagnosing Retrieval-State Lock-In in RAG
Retrieval-state lock-in causes zero-dispersion errors in 42% of KG-RAG and 59% of dense-retrieval failures; a three-object check rule reaches 91.9% pooled precision at 7.7% coverage.
-
Retrievals Can Be Detrimental: Unveiling the Backdoor Vulnerability of Retrieval-Augmented Diffusion Models
BadRDM is a backdoor attack on retrieval-augmented diffusion models that poisons the retrieval database with toxicity surrogates and uses multimodal contrastive learning to force toxic generations from text triggers while preserving benign performance.
-
Healthier LLMs: Retrieval-Augmented Generation for Public Health Question Answering
Hybrid RAG over UK public health guidance sharply raises MCQA accuracy and free-form faithfulness, letting smaller open models match larger closed models without retrieval.
-
Evaluating RAG Reliability under Clean, Misleading, and Mixed Retrieval
Proposes an evaluation framework using parametric override and confidence metrics to assess RAG robustness to clean, poisoned, and mixed retrieval evidence on factoid questions.
-
Securing Retrieval-Augmented Generation: A Taxonomy of Attacks, Defenses, and Future Directions
SLOT organizes RAG security literature by attack Surface, defense Layer, CIA Objective, and Target scope, exposing mismatches between attacks and defenses along a six-stage knowledge pipeline.
-
MemOS: A Memory OS for AI System
MemOS introduces a unified memory management framework for LLMs using MemCubes to handle and evolve different memory types for improved controllability and evolvability.
-
AstroRAG -- A Pagerank-Based Retrieval-Augmented Generation Pipeline for Question Answering in Astronomy
AstroRAG adds PageRank re-ranking after MMR retrieval inside transient per-instance indexes to improve LLM answers on astronomy questions, reaching 79.49 percent accuracy and F1 on AstroQA with Mistral-7B.
-
Reducing Hallucination in Enterprise AI Workflows via Hybrid Utility Minimum Bayes Risk (HUMBR)
HUMBR reduces LLM hallucinations in enterprise workflows by using a hybrid semantic-lexical utility within minimum Bayes risk decoding to identify consensus outputs, with derived error bounds and reported outperformance over self-consistency on benchmarks and production data.
-
Security and Privacy in Retrieval-Augmented Generation: Architectures, Threats, Defenses, and Future Directions for Building Trustworthy Systems
A survey of architectures, threats, defenses, and future directions for security and privacy in RAG systems.