mlsanityData to 14 Oct 2026

Representation-Aware Unlearning via Activation Signatures: From Suppression to Entity-Signature Erasure

arXiv 2601.10566 ↗First posted 15 Jan 2026cs.CL cs.LGcited by 0

On the map: LLM safety and security → Machine unlearning

Abstract (arXiv)

Entity-level unlearning is usually evaluated by what a model says: whether it stops naming the target, refuses a query, or shifts a Truth Ratio distribution. These output-level tests, however, do not show whether a subject's internal representation has been attenuated. We introduce the Entity Representation Unlearning Framework (ERUF), a representation-aware framework that mines subject-specific activation signatures, suppresses the corresponding activation direction, and distills the behavior into LoRA parameters. Among evaluated baselines, ERUF is the only method that jointly achieves surface-level suppression, internal attenuation, and utility preservation. On TOFU forget10, ERUF achieves FQ = 0.99 and MU = 0.62, matching reported oracle utility while approaching oracle forget quality. Across most standard foundation-model settings, ERUF maintains low leakage and low internal target activation, with SMR between 0.00% and 1.10%, EL10 below 0.06, and utility drift below 3%. On Llama-3.1-8B, adversarial entity recovery falls from 63.89% to 20.15%, while name-agnostic recovery decreases by 72.7% to 77.4%. Joint surface/internal diagnostics further reveal scale-dependent behavior in reasoning-prior models that surface metrics alone would miss. We interpret these results as operational evidence of representation-level attenuation, not as a formal guarantee of irreversible deletion.

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 34 of its 51 references that are in the map. First posted on arXiv 15 Jan 2026.

Builds on

34 of 51 references.

Closest ancestors · 10

Same direction, newest first

  1. 2025Unlearning vs. Obfuscation: Are We Truly Removing Knowledge?5 May
  2. 2025LLM Unlearning via Neural Activation Redirection11 Feb
  3. 2024Do Unlearning Methods Remove Information from Language Model Weights?11 Oct
  4. 2024Simplicity Prevails: Rethinking Negative Preference Optimization for LLM Unlearning9 Oct
  5. 2024Position: LLM Unlearning Benchmarks are Weak Measures of Progress3 Oct
  6. 2024MUSE: Machine Unlearning Six-Way Evaluation for Language Models8 Jul
  7. 2024Unveiling Entity-Level Unlearning for Large Language Models: A Comprehensive Analysis22 Jun
  8. 2024Negative Preference Optimization: From Catastrophic Collapse to Effective Unlearning8 Apr
  9. 2024TOFU: A Task of Fictitious Unlearning for LLMs11 Jan
  10. 2022Knowledge Unlearning for Mitigating Privacy Risks in Language Models4 Oct

Other references · 24

  1. 2017Proximal Policy Optimization Algorithmscited by 15,215
  2. 2021LoRA: Low-Rank Adaptation of Large Language Modelscited by 14,380
  3. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 8,065
  4. 2018Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challengecited by 4,886
  5. 2016Overcoming catastrophic forgetting in neural networkscited by 4,787
  6. 2019HellaSwag: Can a Machine Really Finish Your Sentence?cited by 4,100
  7. 2023Universal and Transferable Adversarial Attacks on Aligned Language Modelscited by 3,104
  8. 2021TruthfulQA: Measuring How Models Mimic Human Falsehoodscited by 2,975
  9. 2019PIQA: Reasoning about Physical Commonsense in Natural Languagecited by 2,883
  10. 2022Locating and Editing Factual Associations in GPTcited by 2,691
  11. 2019BoolQ: Exploring the Surprising Difficulty of Natural Yes/No Questionscited by 2,183
  12. 2018Can a Suit of Armor Conduct Electricity? A New Dataset for Open Book Question Answeringcited by 2,164
12 more
  1. 2023Jailbroken: How Does LLM Safety Training Fail?cited by 1,651
  2. 2020Transformer Feed-Forward Layers Are Key-Value Memoriescited by 1,336
  3. 2019Energy and Policy Considerations for Deep Learning in NLPcited by 1,254
  4. 2023Steering Llama 2 via Contrastive Activation Additioncited by 1,085
  5. 2019Machine Unlearningcited by 1,061
  6. 2021Ethical and social risks of harm from Language Modelscited by 966
  7. 2021Carbon Emissions and Large Neural Network Trainingcited by 567
  8. 2024The WMDP Benchmark: Measuring and Reducing Malicious Use With Unlearningcited by 509
  9. 2019Neural Text Generation with Unlikelihood Trainingcited by 503
  10. 2024Great, Now Write an Article About That: The Crescendo Multi-Turn LLM Jailbreak Attackcited by 390
  11. 2023Can Sensitive Information Be Deleted From LLMs? Objectives for Defending Against Extraction Attackscited by 158
  12. 2024Activation Scaling for Steering and Interpreting Language Modelscited by 35

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026Auditing Language Model Unlearning via Information Decomposition
  2. 2024Opt-Out: Investigating Entity-Level Unlearning for Large Language Models via Optimal Transport
  3. 2025Are We Truly Forgetting? A Critical Re-examination of Machine Unlearning Evaluation Protocols
  4. 2026Robust LLM Unlearning Against Relearning Attacks: The Minor Components in Representations Matter
  5. 2026Measuring the Depth of LLM Unlearning via Activation Patching
  6. 2024Challenging Forgets: Unveiling the Worst-Case Forget Sets in Machine Unlearning
  7. 2026Auditing Forgetting in Limited Memory Language Models
  8. 2026Forget Narrowly, Retain Broadly: Unlearning as an Asymmetric Generalization Problem