Robust Utility-Preserving Text Anonymization Based on Large Language Models

TL;DR

This paper introduces a LLM-based framework for text anonymization that balances privacy preservation against re-identification risks and maintains data utility for downstream tasks, with extensive experiments validating its effectiveness.

Contribution

It proposes a novel LLM-driven anonymization framework with dedicated privacy and utility evaluators, and explores distillation for real-time applications, outperforming existing methods.

Findings

Outperforms baseline models in re-identification risk reduction

Preserves higher data utility in downstream tasks

Effective in large-scale, real-time scenarios

Abstract

Anonymizing text that contains sensitive information is crucial for a wide range of applications. Existing techniques face the emerging challenges of the re-identification ability of large language models (LLMs), which have shown advanced capability in memorizing detailed information and reasoning over dispersed pieces of patterns to draw conclusions. When defending against LLM-based re-identification, anonymization could jeopardize the utility of the resulting anonymized data in downstream tasks. In general, the interaction between anonymization and data utility requires a deeper understanding within the context of LLMs. In this paper, we propose a framework composed of three key LLM-based components: a privacy evaluator, a utility evaluator, and an optimization component, which work collaboratively to perform anonymization. Extensive experiments demonstrate that the proposed model outperforms existing baselines, showing robustness in reducing the risk of re-identification while preserving greater data utility in downstream tasks. We provide detailed studies on these core modules. To consider large-scale and real-time applications, we investigate the distillation of the anonymization capabilities into lightweight models. All of our code and datasets will be made publicly available at https://github.com/UKPLab/acl2025-rupta.