OMB M-25-21.
OMB M-25-21 (issued 2025-04-03) is the operative federal memo governing how agencies use AI. The evidence record proves: Produces the reproducible pre-deployment test artifact the testing practice calls for
Reviewed 2026-07-08
What this is.
Use. Federal AI use · Chief AI Officer · high-impact minimum practices (issued 2025-04-03).
Register. 09 · Compliance controls — one of the 17 registers of the PLANiSPHERE corpus library.
See it work.
Planisphere measures your tool deployment against OMB M-25-21 and seals the result into a signed, Merkle-rooted evidence record. The grade recomputes on your own hardware; the model state never crosses the boundary.
Promise. compliance
OMB M-25-21 is a binding federal directive — a thing an agency or contractor MUST do, on a clock. Planisphere produces the pre-deployment and ongoing-monitoring artifacts the directive demands, each pinned to the paragraph that compels it.
Answers: What does OMB M-25-21 require, and what evidence can support review?
What it requires.
OMB M-25-21 (issued 2025-04-03) is the operative federal memo governing how agencies use AI. It rescinded M-24-10, installed Chief AI Officers and AI Governance Boards, defined the high-impact AI category, and attached seven minimum risk-management practices to it — including pre-deployment testing, an AI impact assessment, and ongoing monitoring. Agencies must report their minimum practices to OMB by 2026-09-22, and each of those three practices is operationally a measurement-evidence requirement.
What the evidence record proves.
- Produces the reproducible pre-deployment test artifact the testing practice calls for
- Supplies the ongoing-monitoring drift record via the re-run cadence
- Serves as behavioural-evidence input to the agency's AI impact assessment
- Anchors the high-impact use case to a sha-pinned, reproducible measurement record
What this does not prove. M-25-21's human-oversight, remedies/appeals, and feedback practices are organizational and procedural; Planisphere measures model behaviour, not an agency's governance process. The record is evidence the agency cites — it does not author the impact assessment or stand in for the Chief AI Officer's sign-off.
Cite-anchor: OMB Memorandum M-25-21 (Accelerating Federal Use of AI) · Minimum risk-management practices for high-impact AI
Questions this answers.
The artifact this directive asks for.
OMB M-25-21 asks for an evidence record: OMB M-25-21 (issued 2025-04-03) is the operative federal memo governing how agencies use AI. Read the related defense AI workflow product context at /defense.
High-Impact AI reg 09 · M-25-21 risk class · 7 minimum practices · the pre-deployment testing obligation OMB M-25-21/22 reg 09 · federal AI risk-management + acquisition OMB M-25-22 reg 09 · Federal AI acquisition · anti-lock-in · performance/risk monitoring clauses EO 14179 reg 09 · Removing Barriers to American Leadership in AI (Jan 2025) ATO reg 09 · Authorization to Operate · program-office citationSibling marks. Compliance controls
For each vertical.
OMB M-25-21 for defense AI workflow
NIST AI RMF · DoW RAI · OMB M-25-21 Open the pageOMB M-25-21 for clinical AI workflow
HHS §1557 · HIPAA · FDA PCCP Open the pageOMB M-25-21 for legal AI workflow
FRE 702 · ABA Model Rules Open the pageOMB M-25-21 for tutor and grader workflow
Title VI · FERPA · §504 / IDEA Open the pageSee the record OMB M-25-21 asks for.
The console shows the evidence; the docs show the endpoints; the briefing shows what the product does and does not claim.