Pith. sign in

REVIEW 3 cited by

Privacy-Aware Decoding: Mitigating Privacy Leakage of Large Language Models in Retrieval-Augmented Generation

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2508.03098 v2 pith:UL7T5KV6 submitted 2025-08-05 cs.CL

Privacy-Aware Decoding: Mitigating Privacy Leakage of Large Language Models in Retrieval-Augmented Generation

classification cs.CL
keywords privacydecodinggenerationnoisesensitiveinformationlanguagelarge
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Retrieval-Augmented Generation (RAG) enhances the factual accuracy of large language models (LLMs) by conditioning outputs on external knowledge sources. However, when retrieval involves private or sensitive data, RAG systems are susceptible to extraction attacks that can leak confidential information through generated responses. We propose Privacy-Aware Decoding (PAD), a lightweight, inference-time defense that adaptively injects calibrated Gaussian noise into token logits during generation. PAD integrates confidence-based screening to selectively protect high-risk tokens, efficient sensitivity estimation to minimize unnecessary noise, and context-aware noise calibration to balance privacy with generation quality. A \renyi Differential Privacy (RDP) accountant rigorously tracks cumulative privacy loss, enabling explicit per-response $(\varepsilon, \delta)$-DP guarantees for sensitive outputs. Unlike prior approaches requiring retraining or corpus-level filtering, PAD is model-agnostic and operates entirely at decoding time with minimal computational overhead. Experiments on three real-world datasets demonstrate that PAD substantially reduces private information leakage while preserving response utility, outperforming existing retrieval- and post-processing-based defenses. Our work takes an important step toward mitigating privacy risks in RAG via decoding strategies, paving the way for universal and scalable privacy solutions in sensitive domains. Our code is available: https://github.com/wang2226/PAD.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 3 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Is External Database Protection Static in Retrieval-Augmented Generation? Rethinking Privacy Preservation under Dynamic Queries

    cs.CR 2026-07 reject novelty 6.0

    A query-aware hierarchical differential privacy method for retrieval-augmented generation reports zero extraction leakage in targeted tests while preserving more answer utility than static baselines.

  2. Graphs Don't Stay Secret: Practical Subgraph Reconstruction Attacks on Defended Graph RAG

    cs.CR 2026-02 conditional novelty 6.0

    GRASP, a closed-box multi-turn prompt attack, reconstructs type-faithful one-hop subgraphs from defended Graph RAG systems, reaching up to 82.9 F1 where prior attacks fail.

  3. RAG Security and Privacy: Formalizing the Threat Model and Attack Surface

    cs.CR 2025-09 conditional novelty 3.0

    A formal RAG threat model is defined with four adversary classes and game-based notions of membership inference, leakage, and poisoning, but the definitions largely restate known concepts and the main DP-based protect...