{"slug":"us-omb-m-25-21-high-impact-ai-practices","id":"AIPT-OBL-0116","permanent_url":"https://aipolicytracker.org/id/AIPT-OBL-0116","title":"Apply minimum risk-management practices to high-impact AI","short_title":"High-impact AI practices","category":"risk_management","summary":"For AI whose output serves as a principal basis for decisions with significant effects on rights, safety or access to services, agencies must complete pre-deployment testing, an AI impact assessment, ongoing monitoring, operator training, human oversight and a mechanism for affected people to seek remedy, or stop using the AI.\n","practical_action":"Vendors should prepare test evidence and documentation that agencies can reuse in impact assessments.","is_binding":true,"applies_from":null,"section":null,"source_reference":"Section 4","official_source_url":"https://www.whitehouse.gov/omb/information-for-agencies/memoranda/","review_status":"verified","confidence_level":"medium","last_verified_at":"2026-09-26","reviewed_by":"Bhaskar Bhatt","actors":["public_authority"],"sectors":["public_services"],"use_cases":["public_services","government_procurement"],"applicability":[{"description":"Federal agencies using high-impact AI; indirectly their vendors.","actors":["public_authority","provider"],"conditions":null}],"evidence_examples":[{"title":"AI impact assessment","description":null,"artifact_type":"report"},{"title":"Pre-deployment test results","description":null,"artifact_type":"report"}],"framework_mappings":[{"framework":"nist_ai_rmf","framework_name":"NIST AI RMF 1.0","reference":"MAP, MEASURE, MANAGE","note":"The memo is aligned with the AI RMF vocabulary.","confidence_level":"medium","is_original":true}],"controls":[{"slug":"accuracy-robustness-security-testing","title":"Accuracy, robustness, fairness and security testing","kind":"technical","relationship":"supports","note":"Pre-deployment test evidence agencies can reuse.","confidence_level":"medium","url":"https://aipolicytracker.org/controls/accuracy-robustness-security-testing"},{"slug":"human-oversight-and-override","title":"Human oversight design and override procedure","kind":"process","relationship":"supports","note":"Human oversight and operator arrangements.","confidence_level":"medium","url":"https://aipolicytracker.org/controls/human-oversight-and-override"},{"slug":"ai-impact-assessment","title":"AI impact and fundamental-rights impact assessment","kind":"process","relationship":"supports","note":"AI impact assessment content.","confidence_level":"medium","url":"https://aipolicytracker.org/controls/ai-impact-assessment"}],"url":"https://aipolicytracker.org/obligations/us-omb-m-25-21-high-impact-ai-practices","policy":{"slug":"us-omb-m-25-21","title":"OMB Memorandum M-25-21: Accelerating Federal Use of AI through Innovation, Governance, and Public Trust","status":"in_force","jurisdiction":"us","url":"https://aipolicytracker.org/policies/us-omb-m-25-21"}}