RefineRAG achieves 90% attack success on NQ by generating toxic seeds then optimizing them via retriever-in-the-loop word refinement, outperforming prior methods on effectiveness and naturalness.
In: Proceedings of the AAAI Conference on Artifi- cialIntelligence.vol.38
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CR 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
RefineRAG: Word-Level Poisoning Attacks via Retriever-Guided Text Refinement
RefineRAG achieves 90% attack success on NQ by generating toxic seeds then optimizing them via retriever-in-the-loop word refinement, outperforming prior methods on effectiveness and naturalness.