Auditability

2 articles

Agent Hooks, Two Ways: The In-Process Callback vs. the Externalized Reference Monitor

AI agent hooks offer two primary architectural patterns: In-Process Callback for local development and Externalized Reference Monitor for enterprise security. The choice between them determines whether hooks serve as developer convenience or a robust control plane for governance and auditability.

Sarah Chen·November 6, 2026

A doctor examines a mysterious, glowing black AI box in a hospital setting, highlighting the ethical concerns and lack of transparency in medical AI.

Perspectives

Black Box Medicine: The Three-Part Test for AI You Can Actually Trust

Medical AI systems often operate as "black boxes," posing risks to patient care and informed consent. This article outlines a three-part test—interpretability, auditability, and ethical governance—essential for building trust in AI used in medicine.

Jordan Adebayo·July 14, 2026