TY  - RPRT
TI  - GraphAttack: Exploiting Representational Blindspots in LLM Safety Mechanisms
AU  - Sinan He
AU  - An Wang
PY  - 2025
UR  - https://arxiv.org/abs/2504.13052
ID  - 2504.13052
ER  -