A reinforcement-learning attack that swaps a few words in a single document pushes it into a black-box RAG system's top-3 results and flips the generated answer around 45-47% of the time.
Title resolution pending
1 Pith paper cite this work, alongside 8 external citations. Polarity classification is still indexing.
1
Pith paper citing it
8
external citations · OpenAlex
fields
cs.IR 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
The Silent Saboteur: Imperceptible Adversarial Attacks against Black-Box Retrieval-Augmented Generation Systems
A reinforcement-learning attack that swaps a few words in a single document pushes it into a black-box RAG system's top-3 results and flips the generated answer around 45-47% of the time.