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

arXiv:2610.14763v1 Announce Type: new Abstract: 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.

Related

• OPD-V: Visual On-Policy Self-Distillation with Modality Balance
• ASRU: Activation Steering Meets Reinforcement Unlearning for Multimodal Large Language Models
• The Many Faces of On-Policy Distillation: Pitfalls, Mechanisms, and Fixes
• Rethinking On-Policy Distillation of Large Language Models II: One Training Example
• Multimodal Unlearning Across Vision, Language, Video, and Audio: Survey of Methods, Datasets, and Benchmarks

Source: arXiv cs.AI | 2026-10-15