paper-with-me

홈 › Papers

Learning to Refuse: Towards Mitigating Privacy Risks in LLMs

2024-07-14 · Zhenhua Liu, Tong Zhu, Chuanyuan Tan, Wenliang Chen

Large language models (LLMs) exhibit remarkable capabilities in understanding and generating natural language. However, these models can inadvertently memorize private information, posing significant privacy risks. This study addresses the challenge of enabling LLMs to protect specific individuals' private data without the need for complete retraining. We propose \return, a Real-world pErsonal daTa UnleaRNing dataset, comprising 2,492 individuals from Wikipedia with associated QA pairs, to evaluate machine unlearning (MU) methods for protecting personal data in a realistic scenario. Additionally, we introduce the Name-Aware Unlearning Framework (NAUF) for Privacy Protection, which enables the model to learn which individuals' information should be protected without affecting its ability to answer questions related to other unrelated individuals. Our extensive experiments demonstrate that NAUF achieves a state-of-the-art average unlearning score, surpassing the best baseline method by 5.65 points, effectively protecting target individuals' personal data while maintaining the model's general capabilities.

📄 PDF Abstract BibTeX arXiv:2407.10058

Code (1)

zhliu0106/learning-to-refuse 공식 구현 pytorch

Tasks

Machine Unlearning

Similar Papers 제목 키워드 기반

Neural Gate: Mitigating Privacy Risks in LVLMs via Neuron-Level Gradient Gating

2026-03-13 · Xiangkui Cao, Jie Zhang, Meina Kan, Shiguang Shan 외 arxiv

Large Vision-Language Models (LVLMs) have shown remarkable potential across a wide array of vision-language tasks, leading to their adoption in critical domains such as finance and healthcare. However, their growing depl…

Beyond Data Privacy: New Privacy Risks for Large Language Models

2025-09-16 · Yuntao Du, Zitao Li, Ninghui Li, Bolin Ding arxiv

Large Language Models (LLMs) have achieved remarkable progress in natural language understanding, reasoning, and autonomous decision-making. However, these advancements have also come with significant privacy concerns. W…

Natural Language Understanding

SoK: Privacy Risks and Mitigations in Retrieval-Augmented Generation Systems

2026-01-07 · Andreea-Elena Bodea, Stephen Meisenbacher, Alexandra Klymenko, Florian Matthes arxiv

The continued promise of Large Language Models (LLMs), particularly in their natural language understanding and generation capabilities, has driven a rapidly increasing interest in identifying and developing LLM use case…

Natural Language Understanding

Safer Policy Compliance with Dynamic Epistemic Fallback

2026-01-30 · Joseph Marvin Imperial, Harish Tayyar Madabushi arxiv

Humans develop a series of cognitive defenses, known as epistemic vigilance, to combat risks of deception and misinformation from everyday interactions. Developing safeguards for LLMs inspired by this mechanism might be …

Beyond Visual Evidence: Revealing and Mitigating Relational Privacy Leakage in Document MLLMs

2026-08-13 · Beining Xu, Hairui Wang, Jiaxin Wang, Changsheng Chen 외 arxiv

While the privacy risks of multimodal large language models (MLLMs) have drawn significant attention, the unique vulnerabilities of domain-specific MLLMs remain largely underexplored. Focusing on document understanding M…

Key Information Extraction