TY  - RPRT
TI  - TRAPSBench: Vision-Language Models Encode but Fail to Express Epistemic Restraint
AU  - Fnu Pramono
AU  - John Cai
AU  - Sourabh Kulkarni
PY  - 2026
UR  - https://arxiv.org/abs/2608.13167
ID  - 2608.13167
ER  -