TY  - RPRT
TI  - Jailbreaking Text-to-Image Models Through Cracks: Navigating Heterogeneous Safety Filters via Multi-Agent Debate
AU  - Kaiyan Wen
AU  - Shijie Zhang
AU  - Lu Yu
AU  - Guangdong Bai
PY  - 2026
UR  - https://arxiv.org/abs/2609.01168
ID  - 2609.01168
ER  -