Mask-Free Privacy Extraction and Rewriting: A Domain-Aware Approach via Prototype Learning

TL;DR

DAMPER is a novel domain-aware, mask-free privacy rewriting method for LLMs that uses contrastive learning and prototypes to improve privacy-utility balance with differential privacy guarantees.

Contribution

It introduces a prototype-based approach for autonomous span localization and a preference alignment mechanism for domain-compliant rewriting without human annotations.

Findings

DAMPER outperforms existing methods in privacy-utility trade-offs.

It provides rigorous span-level differential privacy guarantees.

Extensive experiments validate its effectiveness across domains.

Abstract

Client-side privacy rewriting is crucial for deploying LLMs in privacy-sensitive domains. However, existing approaches struggle to balance privacy and utility. Full-text methods often distort context, while span-level approaches rely on impractical manual masks or brittle static dictionaries. Attempts to automate localization via prompt-based LLMs prove unreliable, as they suffer from unstable instruction following that leads to privacy leakage and excessive context scrubbing. To address these limitations, we propose DAMPER (Domain-Aware Mask-free Privacy Extraction and Rewriting). DAMPER operationalizes latent privacy semantics into compact Domain Privacy Prototypes via contrastive learning, enabling precise, autonomous span localization. Furthermore, we introduce a Prototype-Guided Preference Alignment, which leverages learned prototypes as semantic anchors to construct preference pairs, optimizing a domain-compliant rewriting policy without human annotations. At inference time, DAMPER integrates a sampling-based Exponential Mechanism to provide rigorous span-level Differential Privacy (DP) guarantees. Extensive experiments demonstrate that DAMPER significantly outperforms existing baselines, achieving a superior privacy-utility trade-off.