mlsanityData to 14 Oct 2026

CUSE: Fine-Grained Multimodal Unlearning with Counterfactual Guidance and On-Policy Self-Distillation

arXiv 2610.14763 ↗First posted 10 Oct 2026cs.AI cs.CLcited by 0

On the map: Multimodal LLMs / vision-language models → MLLM training and visual instruction tuning

also in 2 more
Abstract (arXiv)

Multimodal large language models (MLLMs) can reproduce sensitive, copyrighted, or unsafe information from their training data. Effective unlearning must remove target information while preserving non-target content, even within the same response. Existing methods typically apply uniform forgetting objectives to complete responses and rely on fixed, off-policy supervision, resulting in coarse token control and misalignment with the model's current behavior. We propose CUSE, a Counterfactual-guided Unlearning framework via on-policy Self-distillation for multimodal large language models. It first constructs counterfactual inputs by removing target-related evidence and uses the resulting changes in token probabilities to identify which tokens require stronger forgetting. It then distills token-level forgetting and retention targets on responses sampled from the current policy, with a frozen copy of the original model serving as the teacher model augmented with the corresponding forgetting- and retention-oriented privileged contexts. This design jointly determines what to forget and what to preserve. On MLLMU-Bench, CUSE reduces forget-set ROUGE-L by 33.6% while achieving 1.2% higher retain-set ROUGE-L than the strongest baseline.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in MLLM training and visual instruction tuning, one of 16 directions in Multimodal LLMs / vision-language models. Builds on 58 of its 61 references that are in the map. First posted on arXiv 10 Oct 2026.

Builds on

58 of 61 references.

Closest ancestors · 4

Same direction, newest first

  1. 2025Qwen3-VL Technical Report26 Nov
  2. 2024Seeing the Image: Prioritizing Visual Correlation by Contrastive Alignment28 May
  3. 2023Improved Baselines with Visual Instruction Tuning5 Oct
  4. 2023MMBench: Is Your Multi-modal Model an All-around Player?12 Jul

Other references · 54

  1. 2015Distilling the Knowledge in a Neural Networkcited by 10,324
  2. 2023BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Modelscited by 6,137
  3. 2025Qwen2.5-VL Technical Reportcited by 5,587
  4. 2023Evaluating Object Hallucination in Large Vision-Language Modelscited by 1,729
  5. 2023On-Policy Distillation of Language Models: Learning from Self-Generated Mistakescited by 939
  6. 2019Eternal Sunshine of the Spotless Net: Selective Forgetting in Deep Networkscited by 618
  7. 2024TOFU: A Task of Fictitious Unlearning for LLMscited by 533
  8. 2024Negative Preference Optimization: From Catastrophic Collapse to Effective Unlearningcited by 504
  9. 2022Knowledge Unlearning for Mitigating Privacy Risks in Language Modelscited by 437
  10. 2023SalUn: Empowering Machine Unlearning via Gradient-based Weight Saliency in Both Image Classification and Generationcited by 386
  11. 2023Who's Harry Potter? Approximate Unlearning in LLMscited by 381
  12. 2024MUSE: Machine Unlearning Six-Way Evaluation for Language Modelscited by 261
38 more
  1. 2024Direct Language Model Alignment from Online AI Feedbackcited by 260
  2. 2023In-Context Unlearning: Language Models as Few Shot Unlearnerscited by 224
  3. 2026On-Policy Context Distillation for Language Modelscited by 186
  4. 2024Simplicity Prevails: Rethinking Negative Preference Optimization for LLM Unlearningcited by 172
  5. 2024Towards Safer Large Language Models through Machine Unlearningcited by 139
  6. 2022Continual Learning and Private Unlearningcited by 130
  7. 2023KGA: A General Machine Unlearning Framework Based on Knowledge Gap Alignmentcited by 99
  8. 2024Guardrail Baselines for Unlearning in LLMscited by 96
  9. 2024Human Alignment of Large Language Models through Online Preference Optimisationcited by 91
  10. 2023DEPN: Detecting and Editing Privacy Neurons in Pretrained Language Modelscited by 79
  11. 2024Protecting Privacy in Multimodal Large Language Models with MLLMU-Benchcited by 64
  12. 2025Black-Box On-Policy Distillation of Large Language Modelscited by 59
  13. 2024UNDIAL: Self-Distillation with Adjusted Logits for Robust Unlearning in Large Language Modelscited by 50
  14. 2024Single Image Unlearning: Efficient Machine Unlearning in Multimodal Large Language Modelscited by 46
  15. 2024Soft Prompting for Unlearning in Large Language Modelscited by 46
  16. 2024Unified Gradient-Based Machine Unlearning with Remain Geometry Enhancementcited by 45
  17. 2024CLEAR: Character Unlearning in Textual and Visual Modalitiescited by 39
  18. 2026Rethinking On-Policy Self-Distillation for Thinking Modelscited by 39
  19. 2024Revisiting Who's Harry Potter: Towards Targeted Unlearning from a Causal Intervention Perspectivecited by 36
  20. 2026GATES: Self-Distillation under Privileged Context with Consensus Gatingcited by 35
  21. 2024Alternate Preference Optimization for Unlearning Factual Knowledge in Large Language Modelscited by 33
  22. 2025SafeEraser: Enhancing Safety in Multimodal Large Language Models through Multimodal Machine Unlearningcited by 32
  23. 2025MMUnlearner: Reformulating Multimodal Machine Unlearning in the Era of Multimodal Large Language Modelscited by 30
  24. 2026Online Experiential Learning for Language Modelscited by 28
  25. 2025Modality-Aware Neuron Pruning for Unlearning in Multimodal Large Language Modelscited by 26
  26. 2024Cross-Modal Safety Alignment: Is textual unlearning all you need?cited by 25
  27. 2024Unified Parameter-Efficient Unlearning for LLMscited by 25
  28. 2024Avoiding Copyright Infringement via Large Language Model Unlearningcited by 24
  29. 2025ReLearn: Unlearning via Learning for Large Language Modelscited by 22
  30. 2024EFUF: Efficient Fine-grained Unlearning Framework for Mitigating Hallucinations in Multimodal Large Language Modelscited by 17
  31. 2025RULE: Reinforcement UnLEarning Achieves Forget-Retain Pareto Optimalitycited by 16
  32. 2025SAUCE: Selective Concept Unlearning in Vision-Language Models with Sparse Autoencoderscited by 13
  33. 2026Reducing the Safety Tax in LLM Safety Alignment with On-Policy Self-Distillationcited by 11
  34. 2025MLLMEraser: Achieving Test-Time Unlearning in Multimodal Large Language Models through Activation Steeringcited by 10
  35. 2025Bridging the Copyright Gap: Do Large Vision-Language Models Recognize and Respect Copyrighted Content?cited by 10
  36. 2025Cross-Modal Unlearning via Influential Neuron Path Editing in Multimodal Large Language Modelscited by 8
  37. 2026Multimodal Unlearning Across Vision, Language, Video, and Audio: Survey of Methods, Datasets, and Benchmarkscited by 7
  38. 2025Towards Reasoning-Preserving Unlearning in Multimodal Large Language Modelscited by 4

50 most cited of 54 All 54 (JSON)

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025Towards Benign Memory Forgetting for Selective Multimodal Large Language Model Unlearning
  2. 2026Toward Fine-Grained Forgetting:Attribute Unlearning for Multimodal Large Language Models
  3. 2026Forgetting Only What Matters: Layer-Selective Unlearning toward Robust LLMs
  4. 2025Not Every Token Needs Forgetting: Selective Unlearning to Limit Change in Utility in Large Language Model Unlearning
  5. 2026Are We Really Measuring Forgetting? Pitfalls in Evaluating Machine Unlearning for Multimodal LLMs
  6. 2025Unlearning That Lasts: Utility-Preserving, Robust, and Almost Irreversible Forgetting in LLMs
  7. 2026Not Every Change Is Necessary: Recoverable Drift in Large Language Model Unlearning
  8. 2026Attention Smoothing Is All You Need For Unlearning