Safety Game: Balancing Safe and Informative Conversations with Blackbox Agentic AI using LP Solvers
This research introduces a novel framework using linear programming to optimize safety and utility in blackbox agentic AI through zero-sum game modeling, off...