Lune

ACL Findings2025

REVS: Unlearning Sensitive Information in Language Models via Rank Editing in the Vocabulary Space

Tomer Ashuach, Martin Tutek, Yonatan Belinkov

Abstract

Language models (LMs) risk inadvertently memorizing and divulging sensitive or personally identifiable information (PII) seen in training data, causing privacy concerns. Current approaches to address this issue involve costly dataset scrubbing, or model filtering through unlearning and model editing, which can be bypassed through extraction attacks. We propose REVS, a novel non-gradient-based method for unlearning sensitive information from LMs. REVS identifies and modifies a small subset of neurons relevant for constituent tokens that form sensitive information. To adequately evaluate our method on truly sensitive information, we curate three datasets: email and URL datasets naturally memorized by the models, and a synthetic social security number dataset that we tune the models to memorize. Compared to other methods, REVS demonstrates superior performance in unlearning sensitive information and robustness to extraction attacks, while retaining underlying model integrity.

Ask Lune

Know this paper in minutes, not an afternoon.

Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.

6 top-tier papers have cited it since. Know what they found before your reviewers do.

Questions to start from

Lune can draw on

the full paper · the top-tier papers that cite it · related work

Get the answer

Free to start. No credit card required.

Lineage

  • Measuring Chain of Thought Faithfulness by Unlearning Reasoning Steps cites Locating and Editing Factual Associations in GPT
  • Measuring Chain of Thought Faithfulness by Unlearning Reasoning Steps cites Precise In-Parameter Concept Erasure in Large Language Models
  • Effective PII Extraction from LLMs through Augmented Few-Shot Learning cites The Secret Sharer: Evaluating and Testing Unintended Memorization in Neural Networks
  • Membership Inference Attacks Against Machine Learning Models cites Deep Learning with Differential Privacy
  • Precise In-Parameter Concept Erasure in Large Language Models cites Locating and Editing Factual Associations in GPT
  • Precise In-Parameter Concept Erasure in Large Language Models cites Measuring Massive Multitask Language Understanding
  • The Secret Sharer: Evaluating and Testing Unintended Memorization in Neural Networks cites Membership Inference Attacks Against Machine Learning Models
  • The Secret Sharer: Evaluating and Testing Unintended Memorization in Neural Networks cites Deep Learning with Differential Privacy
  • The Privacy Paradox of LLMs: User Perceptions and the Reality of PII Leakage cites Effective PII Extraction from LLMs through Augmented Few-Shot Learning
  • The Privacy Paradox of LLMs: User Perceptions and the Reality of PII Leakage cites The Secret Sharer: Evaluating and Testing Unintended Memorization in Neural Networks
  • CRISP: Persistent Concept Unlearning via Sparse Autoencoders cites Measuring Massive Multitask Language Understanding
  • CRISP: Persistent Concept Unlearning via Sparse Autoencoders cites Precise In-Parameter Concept Erasure in Large Language Models
Builds onCited byLeft to right by year; the more cited, the higher.

Cited by top-tier papers6

  1. Measuring Chain of Thought Faithfulness by Unlearning Reasoning StepsMartin Tutek, Fateme Hashemi Chaleshtori, Ana Marasovic, Yonatan BelinkovEMNLP 2025 · 37 citations
  2. Precise In-Parameter Concept Erasure in Large Language ModelsYoav Gur-Arieh, Clara Suslik, Yihuai Hong, Fazl Barez et al.EMNLP 2025 · 10 citations
  3. CRISP: Persistent Concept Unlearning via Sparse AutoencodersTomer Ashuach, Dana Arad, Aaron Mueller, Martin Tutek et al.ACL 2026 · 7 citations
  4. Effective PII Extraction from LLMs through Augmented Few-Shot LearningShuai Cheng, Shu Meng, Haitao Xu, Haoran Zhang et al.USENIX Security 2025
  5. The Privacy Paradox of LLMs: User Perceptions and the Reality of PII LeakageShuai Cheng, Haitao Xu, Shu Meng, Shuai Hao et al.CHI 2026

Builds on32

  1. Measuring Massive Multitask Language UnderstandingDan Hendrycks, Collin Burns, Steven Basart, Andy Zou et al.ICLR 2021 · 7,905 citations
  2. Deep Learning with Differential PrivacyMartín Abadi, Andy Chu, Ian J. Goodfellow, H. Brendan McMahan et al.CCS 2016 · 7,620 citations
  3. Membership Inference Attacks Against Machine Learning ModelsReza Shokri, Marco Stronati, Congzheng Song, Vitaly ShmatikovS&P 2017 · 5,137 citations
  4. Locating and Editing Factual Associations in GPTKevin Meng, David Bau, Alex Andonian, Yonatan BelinkovNeurIPS 2022 · 3,415 citations
  5. The Secret Sharer: Evaluating and Testing Unintended Memorization in Neural NetworksNicholas Carlini, Chang Liu, Úlfar Erlingsson, Jernej Kos et al.USENIX Security 2019 · 1,386 citations

Related papers

  1. Knowledge Unlearning for Mitigating Privacy Risks in Language ModelsJoel Jang, Dongkeun Yoon, Sohee Yang, Sungmin Cha et al.ACL 2023 · 48 citations
  2. Private Memorization Editing: Turning Memorization into a Defense to Strengthen Data Privacy in Large Language ModelsElena Sofia Ruzzetti, Giancarlo A. Xompero, Davide Venditti, Fabio Massimo ZanzottoACL 2025 · 9 citations
  3. Teach LLMs to Phish: Stealing Private Information from Language ModelsAshwinee Panda, Christopher A. Choquette-Choo, Zhengming Zhang, Yaoqing Yang et al.ICLR 2024 · 41 citations
  4. R.R.: Unveiling LLM Training Privacy through Recollection and RankingWenlong Meng, Zhenyuan Guo, Lenan Wu, Chen Gong et al.ACL Findings 2025 · 2 citations
  5. Keeping an Eye on LLM Unlearning: The Hidden Risk and RemedyJie Ren, Zhenwei Dai, Xianfeng Tang, Yue Xing et al.NeurIPS 2025 · 11 citations
Dusk over the sea between two cliffs drawn in fine vertical lines