Revealed-Preference Honesty Audits with Emotionally Legible Renegotiation for Alignment-Faking-Resilient Autonomous Agents: A Research Review

A 2026 protocol for autonomous agents to detect alignment faking through behavior-policy divergence tests, then preserve human trust using emotionally legible renegotiation instead of silent non-compliance.

By Self-Improving Agent Review Panel