How Private are Language Models in Abstractive Summarization?

TL;DR

This paper investigates the privacy risks of language models in abstractive summarization, revealing that current models often leak sensitive information, unlike human summaries which better protect privacy.

Contribution

It provides a comprehensive analysis of privacy risks in LM-based summarization across multiple models and datasets, highlighting the gap with human privacy standards.

Findings

Language models frequently leak personally identifiable information.

Human summaries significantly outperform models in privacy preservation.

There is a substantial gap between current LM privacy capabilities and human performance.

Abstract

In sensitive domains such as medical and legal, protecting sensitive information is critical, with protective laws strictly prohibiting the disclosure of personal data. This poses challenges for sharing valuable data such as medical reports and legal cases summaries. While language models (LMs) have shown strong performance in text summarization, it is still an open question to what extent they can provide privacy-preserving summaries from non-private source documents. In this paper, we perform a comprehensive study of privacy risks in LM-based summarization across two closed- and four open-weight models of different sizes and families. We experiment with both prompting and fine-tuning strategies for privacy-preservation across a range of summarization datasets including medical and legal domains. Our quantitative and qualitative analysis, including human evaluation, shows that LMs frequently leak personally identifiable information in their summaries, in contrast to human-generated privacy-preserving summaries, which demonstrate significantly higher privacy protection levels. These findings highlight a substantial gap between current LM capabilities and expert human expert performance in privacy-sensitive summarization tasks.