Chief Risk Officers and Heads of Model Risk at banks, insurers and specialty lenders.
Yes, and most of our validations are exactly that. Independence is the point. A separate validation pod challenges a model built by your team or a third party, which is the arrangement SR 11-7 and SS1/23 require.
The framework still applies; the test battery changes. We map conceptual soundness, benchmarking and outcomes analysis onto generative behaviour and add prompt-injection, hallucination and drift testing that classic statistical validation never contemplated.