Are Large Pre-Trained Language Models Leaking Your Personal Information?
Abstract
Are Large Pre-Trained Language Models Leaking Your Personal Information? In this paper, we analyze whether Pre-Trained Language Models (PLMs) are prone to leaking personal information. Specifically, we query PLMs for email addresses with contexts of the email address or prompts containing the owner’s name. We find that PLMs do leak personal information due to memorization. However, since the models are weak at association, the risk of specific personal information being extracted by attackers is low. We hope this work could help the community to better understand the privacy risk of PLMs and bring new insights to make PLMs safe.
Cite
Text
Huang et al. "Are Large Pre-Trained Language Models Leaking Your Personal Information?." ICML 2022 Workshops: KRLM, 2022.Markdown
[Huang et al. "Are Large Pre-Trained Language Models Leaking Your Personal Information?." ICML 2022 Workshops: KRLM, 2022.](https://mlanthology.org/icmlw/2022/huang2022icmlw-large/)BibTeX
@inproceedings{huang2022icmlw-large,
title = {{Are Large Pre-Trained Language Models Leaking Your Personal Information?}},
author = {Huang, Jie and Shao, Hanyin and Chang, Kevin},
booktitle = {ICML 2022 Workshops: KRLM},
year = {2022},
url = {https://mlanthology.org/icmlw/2022/huang2022icmlw-large/}
}