Who authorized the algorithm? Reckoning with ungoverned AI

That is what a Fortune 500 CIO I advise discovered last quarter when autonomous AI agents deployed by separate teams accessed customer databases, initiated vendor negotiations and generated compliance reports without a single human sign-off. Nobody verified the context protocols connecting those agents to enterprise systems. Nobody asked whether the AI’s decisions aligned with the company’s risk appetite.