Precise Shield: Explaining and Aligning VLLM Safety via Neuron-Level Guidance

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Shi, Enyi, Shen, Fei, Miao, Shuyi, Zhu, Linxia, Shao, Pengyang, Tang, Jinhui, Chua, Tat-Seng
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917398435594240
author Shi, Enyi
Shen, Fei
Miao, Shuyi
Zhu, Linxia
Shao, Pengyang
Tang, Jinhui
Chua, Tat-Seng
author_facet Shi, Enyi
Shen, Fei
Miao, Shuyi
Zhu, Linxia
Shao, Pengyang
Tang, Jinhui
Chua, Tat-Seng
contents In real-world deployments, Vision-Language Large Models (VLLMs) face critical challenges from multilingual and multimodal composite attacks: harmful images paired with low-resource language texts can easily bypass defenses designed for high-resource language scenarios, exposing structural blind spots in current cross-lingual and cross-modal safety methods. This raises a mechanistic question: where is safety capability instantiated within the model, and how is it distributed across languages and modalities? Prior studies on pure-text LLMs have identified cross-lingual shared safety neurons, suggesting that safety may be governed by a small subset of critical neurons. Leveraging this insight, we propose Precise Shield, a two-stage framework that first identifies safety neurons by contrasting activation patterns between harmful and benign inputs, and then constrains parameter updates strictly within this subspace via gradient masking with affecting fewer than 0.03% of parameters. This strategy substantially improves safety while preserving multilingual and multimodal generalization. Further analysis reveals a moderate overlap of safety neurons across languages and modalities, enabling zero-shot cross-lingual and cross-modal transfer of safety capabilities, and offering a new direction for neuron-level, transfer-based safety enhancement.
format Preprint
id arxiv_https___arxiv_org_abs_2604_08881
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Precise Shield: Explaining and Aligning VLLM Safety via Neuron-Level Guidance
Shi, Enyi
Shen, Fei
Miao, Shuyi
Zhu, Linxia
Shao, Pengyang
Tang, Jinhui
Chua, Tat-Seng
Computer Vision and Pattern Recognition
In real-world deployments, Vision-Language Large Models (VLLMs) face critical challenges from multilingual and multimodal composite attacks: harmful images paired with low-resource language texts can easily bypass defenses designed for high-resource language scenarios, exposing structural blind spots in current cross-lingual and cross-modal safety methods. This raises a mechanistic question: where is safety capability instantiated within the model, and how is it distributed across languages and modalities? Prior studies on pure-text LLMs have identified cross-lingual shared safety neurons, suggesting that safety may be governed by a small subset of critical neurons. Leveraging this insight, we propose Precise Shield, a two-stage framework that first identifies safety neurons by contrasting activation patterns between harmful and benign inputs, and then constrains parameter updates strictly within this subspace via gradient masking with affecting fewer than 0.03% of parameters. This strategy substantially improves safety while preserving multilingual and multimodal generalization. Further analysis reveals a moderate overlap of safety neurons across languages and modalities, enabling zero-shot cross-lingual and cross-modal transfer of safety capabilities, and offering a new direction for neuron-level, transfer-based safety enhancement.
title Precise Shield: Explaining and Aligning VLLM Safety via Neuron-Level Guidance
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2604.08881