IncogniText: Privacy-Enhancing Conditional Text Anonymization via LLM-Based Private Attribute Randomization
Abstract
In this work, we address the problem of text anonymization where the goal is to prevent adversaries from correctly inferring private attributes of the author, while keeping the text utility, i.e., meaning and semantics. We propose IncogniText, a technique that anonymizes the text to mislead a potential adversary into predicting a wrong private attribute value. Our empirical evaluation shows a reduction of private attribute leakage by more than 90% across 8 different private attributes. Finally, we demonstrate the maturity of IncogniText for real-world applications by distilling its anonymization capability into a set of LoRA parameters associated with an on-device model. Our results show the possibility of reducing privacy leakage by more than half with limited impact on utility.
Cite
Text
Frikha et al. "IncogniText: Privacy-Enhancing Conditional Text Anonymization via LLM-Based Private Attribute Randomization." NeurIPS 2024 Workshops: SafeGenAi, 2024.Markdown
[Frikha et al. "IncogniText: Privacy-Enhancing Conditional Text Anonymization via LLM-Based Private Attribute Randomization." NeurIPS 2024 Workshops: SafeGenAi, 2024.](https://mlanthology.org/neuripsw/2024/frikha2024neuripsw-incognitext/)BibTeX
@inproceedings{frikha2024neuripsw-incognitext,
title = {{IncogniText: Privacy-Enhancing Conditional Text Anonymization via LLM-Based Private Attribute Randomization}},
author = {Frikha, Ahmed and Walha, Nassim and Nakka, Krishna Kanth and Mendes, Ricardo and Jiang, Xue and Zhou, Xuebing},
booktitle = {NeurIPS 2024 Workshops: SafeGenAi},
year = {2024},
url = {https://mlanthology.org/neuripsw/2024/frikha2024neuripsw-incognitext/}
}