Definition
The disconnect between an organisation saying it has responsible AI policies and actually measuring whether those policies work in practice. Research consistently shows that most executives claim to pursue 'trusted AI' while only a small fraction have real metrics to prove it.
Why it matters
A governance programme without measurement is unauditable and unenforceable. Boards that accept claims of AI trustworthiness at face value—without asking for quantitative evidence—are exposed to regulatory, legal, and reputational risk when something goes wrong.