Metraipolis › archive › dc_65eb1512ab75

Prevented is Not Protected: What Attack Success Rate Hides about RAG — arXiv:2610.40742

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.40742
License (as archived)
CC0-1.0
Archived
2026-11-03T07:55:21.499Z
SHA-256 of the archived text
322029b27a5470f6046ba9ff4e744b257d49a98562e85089a061304e240dedb4
Stable URL of this page
https://metraipolis.com/documento/dc_65eb1512ab75

How to cite

Prevented is Not Protected: What Attack Success Rate Hides about RAG — arXiv:2610.40742 — arXiv (en), https://arxiv.org/abs/2610.40742 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_65eb1512ab75 (Metraipolis archive dc_65eb1512ab75, sha256 322029b27a5470f6046ba9ff4e744b257d49a98562e85089a061304e240dedb4)
BibTeX
@misc{dc65eb1512ab75,
  title = {Prevented is Not Protected: What Attack Success Rate Hides about RAG — arXiv:2610.40742},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.40742},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_65eb1512ab75; sha256 322029b27a5470f6046ba9ff4e744b257d49a98562e85089a061304e240dedb4},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.40742. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Retrieval-Augmented Generation (RAG) is vulnerable to knowledge poisoning, where adversaries manipulate retrieved evidence to influence model outputs. Existing defenses are primarily evaluated using attack success rate (ASR), which obscures defense-induced failures and the mechanisms behind apparent attack prevention. To address this issue, this work introduces \textbf{Aligned RAG Evaluation}, a matched-query protocol that jointly examines clean and poisoned responses, with and without defense. By combining matched conditions analysis, LLM-judge endorsement, and retrieval dependence controls, the proposed framework distinguishes prevented from introduced failures, target mentions from LLM endorsement, and single-passage influence from evidence composition. Experiments on HotpotQA and Natural Questions reveal that existing defenses can introduce unexpected target-positive answers, although they can prevent some attacks, while evaluation criteria can also change defense rankings. Our findings expose critical blind spots in ASR-based evaluation and motivate a more comprehensive assessment of RAG poisoning defenses. The code and the evaluation artifact are available at this https URL .

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.CL, nuovi lavori del giorno).
Categorie arXiv: Cryptography and Security (cs.CR) ; Computation and Language (cs.CL); Information Retrieval (cs.IR).
Pagina del lavoro: https://arxiv.org/abs/2610.40742 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-11-03T07:49:46.018Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).