works on
large vision-language models 1multilingual safety 1multimodal alignment 1neuron-level alignment 1parameter-efficient fine-tuning 1

From the 1 of 12 linked papers with an AI index.

activity
20242026
collaborators
Showing cs.AIShow all

1 paper · 1 filter

cs.AI2026

One Anchor for All: Unified Multilingual and Multimodal Safety Alignment for LVLMs

Enyi Shi, Fei Shen, Chuancheng Shi +4

The paper introduces a neuron‑level safety alignment method that identifies and updates a tiny set of shared safety neurons across languages and modalities, enabling large vision‑l…