Scholar Feed

Retrieval-augmented generation

TrustRAG

TrustRAG: Enhancing Robustness and Trustworthiness in Retrieval-Augmented Generation

Superseded baseline#29 of 1,179 most-superseded · first seen Jan 1, 2025

Superseded — cited as a baseline and beaten by newer methods

4 papers critique it · 4 beat it on benchmarks

What papers say

Verbatim critique sentences, each from a paper that cites TrustRAG as a baseline.

Existing defenses treat poisoning as a content-quality problem: filtering bad documents ragdefender, detecting anomalous signals revprag,avfilter, scoring trustworthiness trustrag, or isolating passages robustrag. These approaches share an implicit assumption: if the system can identify poisoned evidence, it will naturally avoid acting on it. We show this assumption is incorrect.
— Cordon-MAS: Defending RAG against Knowledge Poisoning via Information-Flow Control
“it makes the unrealistic assumption that malicious documents form a separate cluster in the embedding space.”
— ReliabilityRAG: Effective and Provably Robust Defense for RAG-based Web-Search
“For numerical manipulation: the poisoned passage clusters with the legitimate passage (cosine similarity 0.9997 means they are in the same cluster). Stage 1 clustering cannot separate them. Stage 2 LLM self-assessment would need to notice that `$15,500` ≠ `$15,000` across two passages in the context window, but research on LLM numerical reasoning shows this is unreliable, especially when the numbers are embedded in otherwise identical text.”
— RAGShield: Provenance-Verified Defense-in-Depth Against Knowledge Base Poisoning in Government Retrieval-Augmented Generation Systems
“Although promising, these approaches have two major limitations: Majority-voting often fails under high poisoning, while heuristic and aggressive filtering may lose relevant content under low poisoning.”
— SeCon-RAG: A Two-Stage Semantic Filtering and Conflict-Free Framework for Trustworthy RAG

Beaten on benchmarks

Head-to-head results where a newer method reports beating TrustRAG. Values are copied from the source paper's tables — verify against the cited paper.

What to use instead

Recent methods in the same sub-problem, not yet superseded in the knowledge base — arXiv benchmark leaders, not vetted production recommendations.