mlsanityData to 14 Oct 2026

From Anchors to Supervision: Memory-Graph Guided Corpus-Free Unlearning for Large Language Models

arXiv 2604.13777 ↗First posted 15 Apr 2026cs.CL cs.AIcited by 1

On the map: LLM safety and security → Machine unlearning

Abstract (arXiv)

Large language models (LLMs) may memorize sensitive or copyrighted content, raising significant privacy and legal concerns. While machine unlearning has emerged as a potential remedy, prevailing paradigms rely on user-provided forget sets, making unlearning requests difficult to audit and exposing systems to secondary leakage and malicious abuse. We propose MAGE, a Memory-grAph Guided Erasure framework for user-minimized, corpus-free unlearning. Given only a lightweight user anchor that identifies a target entity, MAGE probes the target LLM to recover target-related memorization, organizes it into a weighted local memory graph, and synthesizes scoped supervision for unlearning. MAGE is model-agnostic, can be plugged into standard unlearning methods, and requires no access to the original training corpus. Experiments on two benchmarks, TOFU and RWKU, demonstrate that MAGE's self-generated supervision achieves effective unlearning performance comparable to supervision generated with external reference, while preserving overall utility. These results support a practical and auditable unlearning workflow driven by minimal anchors rather than user-supplied forget corpora.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Placed in Machine unlearning, one of 12 directions in LLM safety and security. Builds on 21 of its 29 references that are in the map. First posted on arXiv 15 Apr 2026.

Builds on

21 of 29 references.

Closest ancestors · 14

Same direction, newest first

  1. 2025Keeping an Eye on LLM Unlearning: The Hidden Risk and Remedy31 May
  2. 2025A General Framework to Enhance Fine-tuning-based LLM Unlearning25 Feb
  3. 2025ReLearn: Unlearning via Learning for Large Language Models16 Feb
  4. 2024Does Unlearning Truly Unlearn? A Black Box Evaluation of LLM Unlearning Methods18 Nov
  5. 2024Position: LLM Unlearning Benchmarks are Weak Measures of Progress3 Oct
  6. 2024Unveiling Entity-Level Unlearning for Large Language Models: A Comprehensive Analysis22 Jun
  7. 2024Negative Preference Optimization: From Catastrophic Collapse to Effective Unlearning8 Apr
  8. 2024Eight Methods to Evaluate Robust Unlearning in LLMs26 Feb
  9. 2024Towards Safer Large Language Models through Machine Unlearning15 Feb
  10. 2024Selective Forgetting: Advancing Machine Unlearning Techniques and Evaluation in Language Models8 Feb
  11. 2024TOFU: A Task of Fictitious Unlearning for LLMs11 Jan
  12. 2023In-Context Unlearning: Language Models as Few Shot Unlearners11 Oct
2 more
  1. 2023Who's Harry Potter? Approximate Unlearning in LLMs3 Oct
  2. 2022Knowledge Unlearning for Mitigating Privacy Risks in Language Models4 Oct

Other references · 7

  1. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 8,065
  2. 2020Extracting Training Data from Large Language Modelscited by 2,189
  3. 2023Can Sensitive Information Be Deleted From LLMs? Objectives for Defending Against Extraction Attackscited by 158
  4. 2025Tokens for Learning, Tokens for Unlearning: Mitigating Membership Inference Attacks in Large Language Models via Dual-Purpose Trainingcited by 15
  5. 2025Proactive Privacy Amnesia for Large Language Models: Safeguarding PII with Negligible Impact on Model Utilitycited by 10
  6. 2025Forgetting to Forget: Attention Sink as A Gateway for Backdooring LLM Unlearningcited by 6
  7. 2025When Forgetting Triggers Backdoors: A Clean Unlearning Attackcited by 2

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026FSFM: A Biologically-Inspired Framework for Selective Forgetting of Agent Memorycited by 3

Similar, not cited

Close in wording, no citation either way

  1. 2024A Closer Look at Machine Unlearning for Large Language Models
  2. 2025Concept Unlearning in Large Language Models via Self-Constructed Knowledge Triplets
  3. 2026Unlearning in LLMs: Methods, Evaluation, and Open Challenges
  4. 2024To Each (Textual Sequence) Its Own: Improving Memorized-Data Unlearning in Large Language Models
  5. 2026Graph-Guided Selective Unlearning for Language Models: Controlling Support Routes Beyond Forget Seeds
  6. 2025Scalable and Robust LLM Unlearning by Correcting Responses with Retrieved Exclusions
  7. 2025Hierarchical Federated Unlearning for Large Language Models
  8. 2025GUARD: Guided Unlearning and Retention via Data Attribution for Large Language Models