Where Identity Lives: Localized, Retain-Free Identity Unlearning in Multimodal Large Language Models

arXiv:2608.30649 · cs.CL, cs.CV · Submitted 2026-08-31 · Read on arXiv

cs.CL, cs.CV

Submitted: 2026-08-31

Updated: 2026-08-31

Comments: Accepted to Findings of the Association for Computational Linguistics: EMNLP 2026

License: http://creativecommons.org/licenses/by/4.0/

The gist: Removing a specific individual's information from multimodal large language models (MLLMs) is often needed after deployment, but existing methods rely on a retain set, which is hardest to obtain at

Terminology

Abstract

Removing a specific individual's information from multimodal large language models (MLLMs) is often needed after deployment, but existing methods rely on a retain set, which is hardest to obtain at that point, and rebuilding it recreates the privacy exposure that unlearning aims to remove. Forgetting from the forget set alone instead damages the shared visual-language computation, harming perception. We cast retain-free unlearning as a localization problem: causal tracing, weight transplant, and Fisher overlap all point to early-to-mid decoder MLPs as the layers where identity information is stored and, unlike other module families, can be modified without substantially disrupting vision. We turn this into Pathway-Aware Visual-attribute Anchoring (PAVA), which confines updates to these layers and pairs a forget loss with a visual-attribute anchor that preserves image-grounded behavior by distilling the model's own pre-unlearning answers from the forget images alone. On MLLMU-Bench and ReMem, PAVA gives the strongest forget-retain trade-off among forget-set-only methods and remains competitive with retain-based baselines.

Sources

Related papers