TY  - RPRT
TI  - When Safe Models Merge into Danger: Exploiting Latent Vulnerabilities in LLM Fusion
AU  - Jiaqing Li
AU  - Zhibo Zhang
AU  - Shide Zhou
AU  - Yuxi Li
AU  - Tianlong Yu
AU  - Kailong Wang
PY  - 2026
UR  - https://arxiv.org/abs/2604.00627
ID  - 2604.00627
ER  -