TY  - RPRT
TI  - SafeCtrl-RL: Inference-Time Adaptive Behaviour Control for LLM Dialogue via RL-Driven Prompt Optimisation
AU  - Michael Orme
AU  - Yanchao Yu
AU  - Zhiyuan Tan
PY  - 2026
UR  - https://arxiv.org/abs/2605.25984
ID  - 2605.25984
ER  -