aiqbench analysis
This research introduces GoldenRetriever, a non-interactive homomorphic encryption framework for Retrieval-Augmented Generation (RAG) that eliminates the high latency of traditional ranking-based privacy methods. By shifting to threshold selection and utilizing CKKS-based computation with precision-stable mask polarization, it reduces computational complexity from quadratic to linear while maintaining retrieval accuracy without exposing query content or intermediate scores. The work establishes a practical foundation for scaling secure RAG systems in environments where data confidentiality is paramount.
GoldenRetriever: Non-Interactive Homomorphic Encrypted Retrieval for Privacy-Preserving RAG
arXiv:2607.29019v1 Announce Type: cross Abstract: Retrieval-Augmented Generation (RAG) enhances large language models by incorporating external knowledge, but existing pipelines typically operate on plaintext data, raising significant privacy concerns. Prior work on privacy-preserving retrieval leverages cryptographic techniques such as homomorphic encryption (HE) and private information retrieval (PIR), but often relies on interactive protocols or ranking-based selection mechanisms that incur high latency and potential information leakage. In this paper, we propose a practical non-interactive encrypted retrieval framework for RAG based on threshold selection. Instead of performing expensive top-$k$ ranking under encryption, our approach selects documents whose similarity scores exceed a predefined threshold, reducing computational complexity from quadratic to linear in the corpus size. We implement this design using CKKS-based homomorphic computation, enabling fully encrypted similarity evaluation and document selection without revealing query content, intermediate scores, or selected indices. To bridge the gap between approximate encrypted computation and discrete token reconstruction, we introduce a precision-stable mask polarization method that ensures accurate recovery of selected documents. Experiments on standard retrieval benchmarks demonstrate that our approach achieves competitive retrieval effectiveness while significantly reducing latency compared to ranking-based encrypted methods. These results highlight threshold-based selection as a practical foundation for scalable and secure RAG systems.
Abstract:Retrieval-Augmented Generation (RAG) enhances large language models by incorporating external knowledge, but existing pipelines typically operate on plaintext data, raising significant privacy concerns. Prior work on privacy-preserving retrieval leverages cryptographic techniques such as homomorphic encryption (HE) and private information retrieval (PIR), but often relies on interactive protocols or ranking-based selection mechanisms that incur high latency and potential information leakage. In this paper, we propose a practical non-interactive encrypted retrieval framework for RAG based on threshold selection. Instead of performing expensive top-$k$ ranking under encryption, our approach selects documents whose similarity scores exceed a predefined threshold, reducing computational complexity from quadratic to linear in the corpus size. We implement this design using CKKS-based homomorphic computation, enabling fully encrypted similarity evaluation and document selection without revealing query content, intermediate scores, or selected indices. To bridge the gap between approximate encrypted computation and discrete token reconstruction, we introduce a precision-stable mask polarization method that ensures accurate recovery of selected documents. Experiments on standard retrieval benchmarks demonstrate that our approach achieves competitive retrieval effectiveness while significantly reducing latency compared to ranking-based encrypted methods. These results highlight threshold-based selection as a practical foundation for scalable and secure RAG systems.
Cryptography and Security (cs.CR); Computation and Language (cs.CL); Information Retrieval (cs.IR)
arXiv:2607.29019 [cs.CR]
(or arXiv:2607.29019v1 [cs.CR] for this version)
https://doi.org/10.48550/arXiv.2607.29019 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 296 words
Related coverage
Balancing Privacy and Efficiency: Music Information Retrieval via Additive Homomorphic Encryption
arXiv · Jul 31, 2026
Privacy-Preserving Text Inference for Health-Misinformation Classification and Semantic Retrieval
arXiv · Oct 16, 2026
End-to-end encrypted ML inference with Amazon SageMaker AI and FHE
AWS · Jun 8, 2026
Dropbox Integrates MCP and Dash to Close the Gap between Security Design and Code Review
InfoQ · Jul 31, 2026