A security team is implementing responsible AI governance controls for a financial services organization deploying predictive AI models. The governance framework requires a human review process for all AI-driven decisions that exceed a certain financial impact threshold. Which Azure capability best supports implementing this human-in-the-loop control?

D is correct because implementing human-in-the-loop review for high-impact AI decisions requires a workflow orchestration solution. Azure Logic Apps can be triggered by AI model outputs that exceed defined thresholds, automatically routing the decision to a human reviewer before finalizing the action. A is incorrect because model monitoring with data drift detection identifies when input data distribution changes, triggering retraining—not human review of individual high-impact decisions. B is incorrect because evaluation flows assess model performance during development, not production decision routing. C is incorrect because the Responsible AI dashboard provides fairness and interpretability analysis during model development, not production-time human review routing.