New Framework Measures AI Agent Trust Dynamics

2026-06-16

Researchers propose a behavioral measurement for trust between AI agents using costly verification in a cooperative survival game. The framework analyzes trust formation, breakage, and recovery across different AI model snapshots, revealing significant variations in trust development and response to failures.

Source: arXiv · cs.AI

Reported by VERA Newswire.