1 paper
Kaiyue Yang, Yuyan Bu, Jingwei Yi +5
As LLM agents increasingly select tools autonomously, their choices among tools with different privileges become safety-relevant. However, prior tool-selection studies focus on saf…