Two frameworks are relevant when artificial intelligence is used in federal statistical work: FCSM 20-04 (A Framework for Data Quality) for evaluating data product quality, and NIST AI RMF 1.0 (Artificial Intelligence Risk Management Framework) for evaluating AI system trustworthiness. Neither references the other, and no published crosswalk connecting them was identified in a literature search. NIST maintains twelve crosswalks to other frameworks, but none from the federal statistical community. Without a shared crosswalk, each agency must independently interpret how the two frameworks relate, producing inconsistent approaches that undermine the coherence both frameworks are designed to provide. A public crosswalk provides a common reference point: a consistent interpretation that practitioners can adopt, evaluate, and refine as both frameworks and operational experience evolve. This article presents the first FCSM × NIST AI RMF crosswalk. Part I maps 11 FCSM dimensions and 7 NIST characteristics across five relationship types: 4 direct mappings where existing practice satisfies both frameworks, 3 partial or split mappings requiring judgment or decomposition, 3 distributed mappings where FCSM functionally addresses NIST concerns across multiple dimensions without naming them, and 4 FCSM-only dimensions outside NIST's scope. A companion appendix provides the complete subcategory-level mapping across all 72 NIST AI RMF subcategories, formalized as a graph database and validated for bidirectional consistency. Part II examines where framework-level mapping reaches its limits, particularly for AI-specific failure modes like confabulation that require domain-specific validation. Part III demonstrates two research implementations showing how the crosswalk tailors to different operational contexts.