1
PsychoSafe: Eliciting Psychologically-Informed Refusals in Large Language Models
用心理学原理优化大模型拒绝回答，减少生硬有害回应，提升AI交互安全性。
arXiv:2606.09697v1 Announce Type: new Abstract: Large language models (LLMs) routinely face requests that should be refused, creating a trade-off betw…