EUVIMEDEUVIMED entdecken
Crossref

Sentinel-Rag: A Controlled Comparison of a Scratch-Trained Classifier and a Local LLM Judge for Defending Enterprise Retrieval-Augmented Generation against Indirect Prompt Injection

Waaridh Borpujari, Aadi Surve, Soham Arte, Sushil Sonawane, Vijaya Patil

International Journal of Advanced Research in Science Communication and Technology · 17.11.2026 · 188

DOI: 10.48175/ijarsct-40823 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Journal of Advanced Research in Science Communication and Technology, online erschienen am 17.11.2026. Verfasst von Waaridh Borpujari und 4 weiteren Autor:innen. Die Arbeit nennt 32 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Enterprise assistants built on retrieval-augmented generation (RAG) read company chats, wikis and policies, so anyone who can post into those sources can plant instructions that the assistant may follow. We present Sentinel-RAG, a controlled testbed in which three assistants share the same corpus, retriever and answer model and differ only in the security stage between retrieval and generation: none, role-based access control plus a 1.6-million-parameter Transformer classifier trained from scratch, or the same access control plus a local 2-billion-parameter LLM judge with schema-constrained output. The classifier reaches 89.8% accuracy and 0.895 macro F1 on held-out attack patterns. On a 25-question suite under two corpus conditions, the unguarded assistant gave 4 and 10 unsafe answers, while both guards gave none. The classifier added 8–9 ms per query but removed 9–16% of benign chunks; the judge removed none but added 13–14 s per query.

Zitieren

APA

Borpujari, W., Surve, A., Arte, S., Sonawane, S., & Patil, V. (2026). Sentinel-Rag: A Controlled Comparison of a Scratch-Trained Classifier and a Local LLM Judge for Defending Enterprise Retrieval-Augmented Generation against Indirect Prompt Injection. International Journal of Advanced Research in Science Communication and Technology, 188. https://doi.org/10.48175/ijarsct-40823

Vancouver

Borpujari W, Surve A, Arte S, Sonawane S, Patil V. Sentinel-Rag: A Controlled Comparison of a Scratch-Trained Classifier and a Local LLM Judge for Defending Enterprise Retrieval-Augmented Generation against Indirect Prompt Injection. International Journal of Advanced Research in Science Communication and Technology. 2026:188. doi:10.48175/ijarsct-40823

BibTeX
@article{Borpujari2026,
  author = {Borpujari, Waaridh and Surve, Aadi and Arte, Soham and Sonawane, Sushil and Patil, Vijaya},
  title = {Sentinel-Rag: A Controlled Comparison of a Scratch-Trained Classifier and a Local LLM Judge for Defending Enterprise Retrieval-Augmented Generation against Indirect Prompt Injection},
  journal = {International Journal of Advanced Research in Science Communication and Technology},
  year = {2026},
  pages = {188},
  doi = {10.48175/ijarsct-40823}
}
Kostenlos starten →