arXiv cs.CL
· Papers
SPADE-Bench: Evaluating Spontaneous Strategic Deception in Agents via Plan-Action Divergence
arXiv:2606.02380v2 Announce Type: replace Abstract: As LLM-based agents expand their operational scope, reliability becomes a prerequisite for real-world deployment. However, in practical applications, human users cannot monitor every immediate behavior; instead, the execution process often remains a black box, leaving