# Decouple to Forget: Semantic Mediation for Sensitive Unlearning in Large Language Models

arXiv: https://arxiv.org/abs/2610.17654 · HTML: https://mlsanity.com/paper/2610.17654/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Machine unlearning, one of 12 directions in LLM safety and security.
- Builds on 23 of its 36 references that are in the map.
- First posted on arXiv 13 Oct 2026.

## Placed in

- LLM safety and security / [Machine unlearning](https://mlsanity.com/safety/machine-unlearning/)

## Builds on (most cited first, 23 of 23)

- [Adam: A Method for Stochastic Optimization](https://mlsanity.com/paper/1412.6980/) (2014)
- [Llama 2: Open Foundation and Fine-Tuned Chat Models](https://mlsanity.com/paper/2307.09288/) (2023)
- [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://mlsanity.com/paper/2305.18290/) (2023)
- [Extracting Training Data from Large Language Models](https://mlsanity.com/paper/2012.07805/) (2020)
- [Jailbroken: How Does LLM Safety Training Fail?](https://mlsanity.com/paper/2307.02483/) (2023)
- [TOFU: A Task of Fictitious Unlearning for LLMs](https://mlsanity.com/paper/2401.06121/) (2024)
- [Negative Preference Optimization: From Catastrophic Collapse to Effective Unlearning](https://mlsanity.com/paper/2404.05868/) (2024)
- [Knowledge Unlearning for Mitigating Privacy Risks in Language Models](https://mlsanity.com/paper/2210.01504/) (2022)
- [Simplicity Prevails: Rethinking Negative Preference Optimization for LLM Unlearning](https://mlsanity.com/paper/2410.07163/) (2024)
- [Memorization vs. Generalization: Quantifying Data Leakage in NLP Performance Evaluation](https://mlsanity.com/paper/2102.01818/) (2021)
- [Rethinking LLM Unlearning Objectives: A Gradient Perspective and Go Beyond](https://mlsanity.com/paper/2502.19301/) (2025)
- [Unlearning as multi-task optimization: A normalized gradient difference approach with an adaptive learning rate](https://mlsanity.com/paper/2410.22086/) (2024)
- [Causal Prompting: Debiasing Large Language Model Prompting based on Front-Door Adjustment](https://mlsanity.com/paper/2403.02738/) (2024)
- [Online Preference Alignment for Language Models via Count-based Exploration](https://mlsanity.com/paper/2501.12735/) (2025)
- [MEOW: MEMOry Supervised LLM Unlearning Via Inverted Facts](https://mlsanity.com/paper/2409.11844/) (2024)
- [Which Retain Set Matters for LLM Unlearning? A Case Study on Entity Unlearning](https://mlsanity.com/paper/2502.11441/) (2025)
- [RULE: Reinforcement UnLEarning Achieves Forget-Retain Pareto Optimality](https://mlsanity.com/paper/2506.07171/) (2025)
- [REVS: Unlearning Sensitive Information in Language Models via Rank Editing in the Vocabulary Space](https://mlsanity.com/paper/2406.09325/) (2024)
- [COPR: Continual Human Preference Learning via Optimal Policy Regularization](https://mlsanity.com/paper/2402.14228/) (2024)
- [Hallucinate Less by Thinking More: Aspect-Based Causal Abstention for Large Language Models](https://mlsanity.com/paper/2511.17170/) (2025)
- [Self-Distilled Disentangled Learning for Counterfactual Prediction](https://mlsanity.com/paper/2406.05855/) (2024)
- [GUARD: Guided Unlearning and Retention via Data Attribution for Large Language Models](https://mlsanity.com/paper/2506.10946/) (2025)
- [Causality-aligned Prompt Learning via Diffusion-based Counterfactual Generation](https://mlsanity.com/paper/2507.19882/) (2025)

## Abstract (arXiv)

Machine unlearning for large language models seeks to remove specified knowledge while preserving general capability, yet current methods often treat an entire response as the unit of forgetting. This coarse granularity causes unnecessary utility loss when the forget target is only a small set of sensitive facts embedded within otherwise benign explanations. In this paper, we introduce Decouple to Forget (D2F), a targeted unlearning method inspired by semantic mediation that separates sensitive factual content from non-sensitive meaning and surface expression. For each example to be forgotten, D2F extracts a fact-level "mediator" representation from the answer, identifies which parts are sensitive, and removes only those parts while leaving the rest of the structure intact. We then generate counterfactual answers grounded in this mediator that preserve non-sensitive content but exclude sensitive facts, employing preference-based contrastive optimization to reduce the likelihood of information leakage while maintaining helpfulness and fluency. Experiments across various benchmark datasets show that D2F achieves a better balance between forgetting and retention than strong baseline methods. Our code is available at https://github.com/JingyueCong/D2F.