Geometric RAG access control isolates policies but not embeddings
A curious user can recover original embeddings from a few hundred pairs and forge policy-compatible queries.
Stories per day · 30 days
11 RSS
Relatedretrieval9benchmarks4agents3embeddings3llm3open-source3privacy3decision-models2differential-privacy2evaluation2poisoning2retrieval-poisoning2access-control1adversarial-queries1agent1agentic-memory1agent-memory1ai-framework1
92 stories, newest first · one card per story
Stories tagged #rag
SNAPRAG extracts RAG knowledge bases with fewer black-box queries
The attack mutates tokens without an external LLM and, the authors say, needs up to 6.77 times fewer queries to extract half a knowledge base.
ArcRAG enables construction-agnostic graph retrieval
Query-local graph interface induction allows fixed retriever to work on differently built knowledge graphs.
Researchers identify text-vector integrity gap in offline RAG
Offline RAG knowledge packs can be hijacked with up to 5.5% of queries exposed due to text-vector mismatch.
IBM patches critical Langflow flaws enabling authentication bypass and RCE
The patches fix four vulnerabilities in Langflow versions 1.0.0 to 1.12.8, including two critical ones rated 9.8 CVSS, and affect Hono framework upgrades.
HERIT adapts translation models for Hanja-to-English historical archives
Authors use RAG on Hanja-Korean corpora to build pseudo-labels, cut temporal bias, and report gains over baselines.
TRIO aligns three agents to refine retrieval-augmented generation
Query, judge, and answer agents are trained with separate rewards, including voting labels and curriculum matching.
Retrieval-augmented generation system for stock market information retrieval
A RAG system for stock market queries achieves high ROUGE scores and 1.16 second average response time on 100 queries.
DeFT-RAG decomposes multi-hop questions for RAG
DeFT-RAG reduces LLM calls by 12.9% through QA cache reuse while achieving top accuracy on multi-hop benchmarks.
Glance reranks passages from the generator's shared prefill
A resident 9B backbone adds about 0.1 GiB per skill and beats bge-reranker-v2-m3 on SciFact and NFCorpus after 39 minutes of fine-tuning.
LEANN open-source vector index shrinks a 60M-chunk RAG index to 6 GB
The only stated figure is a drop from 201 GB to 6 GB; no method or benchmark details were in the feed.
Decision models type RAG source conflicts cheaper than Gemini
Eleven decision models hit 50.0–67.0% on CONFLICTS, and 10 matched or beat Gemini 2.5 Flash at 4–39x lower billed cost.
Decision models classify conflicts in RAG sources
Typed decision models classify knowledge conflicts in RAG faster and cheaper than LLMs while preserving response behavior.
DecisionRAG retrieves passages with token-free decision models
Without task-specific training, it beats dense retrieval plus an 8B reranker by 13.8 R@1 on MultiDoc2Dial.
GuardRAG filters poisoned document clusters after RAG retrieval
The training-free method clusters expanded retrieval results and prunes suspicious groups before they reach the model.
HerHealthGPT grounds reproductive-health answers in retrieval and clarification
On 540 messages from 90 cases, retrieval lifted concern-category accuracy from 54.1% to 59.3%.
Memory Tax argues many agent memory systems lose to naive RAG
The authors split memory's net value into gains over raw-history retrieval and losses they call memory tax.
Paper says RAG poison filters rely on construction shortcuts, not content
Stripping PoisonedRAG's question prefix drops reimplemented RAGuard from AUROC 0.953 to 0.758 and raises its miss rate to 0.763.
Signed knowledge packs still accept poisoned updates in offline edge RAG
On an isolated Raspberry Pi 5, a validly signed poisoned pack drove correct answers on 30 questions to zero.
Evidence-aware retrieval scores reported for a breast-lesion RAG system
The selected hybrid retriever reached 0.94 on Recall@3, R-Precision, MRR@3, and nDCG@3 for 17 of 18 queries.
WD says inference leaves far more data than users see
An 8-second 3.97MB clip left more than 500MB of logs, caches, renders, backups, and safety records.
1
Naive RAG beats complex agentic memory on conversation benchmarks
Raw-RAG ranked first in six of seven settings, ahead of the best agentic system by 0.13 to 5.53 points.
LASAGNA finds RAG component gains often fail to compose
Three techniques predicted a 47% combined gain but delivered 17% together across 450 configurations.
0.5B model detects prompt injection in local RAG pipelines
Authors report 0.994 in-distribution recall at a 0.026 false-positive rate, but near-zero recall on PoisonedRAG.