이 페이지에서3분 읽기
개요
As of the 2025 cycle, OMB M-25-21 directs most agencies to submit at least annual inventories and publish a public version, while law, security limits and reporting criteria determine what is included.
심층 분석
Federal AI inventories are public reporting records, not a complete census of every algorithm or an audit of system performance. Their framework combines Executive Order 13960, the Advancing American AI Act in the FY2023 National Defense Authorization Act, and OMB implementation memoranda. Section 7225 of the Act requires inventories of current and planned agency AI use cases and public availability subject to applicable law and policy; the statute specified a five-year period after enactment. OMB Memorandum M-25-21, issued April 3, 2025, sets an at-least-annual reporting cadence, submission to OMB, and a public version on agency websites. It excludes the Department of Defense and Intelligence Community from that memo’s inventory task. OMB’s 2025 reporting instructions translated these authorities into a data call with individually reported cases and consolidated reporting for some commercial off-the-shelf products. The instructions set a December 22, 2025 submission deadline to OMB and January 28, 2026 public posting date. They also set boundaries: some national-security, intelligence, research and other protected uses are outside reporting scope or not public, while high-impact or operational research cases can still require individual reporting. Exact exemptions and fields depend on current instructions, not the title of a product. The central OMB repository consolidates agency submissions and links to source agency inventories. Its 2025 release describes itself as a compilation of publicly available agency files. Entries are self-reported and reflect the fields and scope of that reporting cycle; a listing does not prove safety, effectiveness or completeness. Conversely, absence from the public repository does not establish that no AI is used. Treat the inventory as a transparency starting point and check the reporting year, agency source and limitations before drawing conclusions.
전략적 영향
위험과 안전
치명적인 AI 피해와 일상적인 AI 피해는 누가 위험을 이해하고 누가 조치를 취할 수 있는지에 따라 달라집니다.
더 명확한 결정들
공공 및 전문 지식은 강력한 안전 정책이 정치적으로 가능한지 여부를 결정합니다.
과장된 과장을 뚫고 나가기
명확한 설명은 과대광고, 연구실 홍보, 모호한 윤리 연극에 의한 포착을 줄입니다.
The Future of Federal Agency AI Use Case Inventories
OMB’s M-25-21 and the 2025 instructions establish the latest cycle documented in the public central repository as of April 2026. Future cycles may update fields, exemptions, deadlines and consolidation rules. The inventory statute’s stated five-year period and later agency guidance are separate legal sources, so check both when describing long-term authority. Agencies should verify new data-call instructions, watch for revised authorities, and note what each reporting cycle can and cannot tell the public. Do not assume a later publication date until OMB or the responsible agency posts it.
실제 구현
A department records a planned tool that will triage applications, identifying its purpose, lifecycle stage, affected users and whether it is high-impact.
A component checks OMB’s current instructions before deciding whether a research project, commercial software feature or national-security use must be reported individually.
A public researcher uses OMB’s consolidated inventory to find a case, then follows the linked agency inventory for fuller context.
An agency excludes a protected detail from its public file but reports releasable fields and submits the information required to OMB.
위험 및 가드레일
실존적 위험을 공상과학처럼 다루면서 능력을 합성합니다.
높은 자율성 하에서 정렬과 표면 제품 안전성을 혼동합니다.
영어가 아니거나 전문가가 아닌 청중에게는 품질이 낮은 소스만 남겨 둡니다.
구현 로드맵
제품 손상, 오용, 통제력 상실/잘못 정렬 위험을 분리합니다.
일정과 심각도에 대한 귀하의 견해를 바꿀 수 있는 증거가 무엇인지 물어보십시오.
마케팅 주장보다 기본 소스와 구체적인 평가를 선호하세요.
인식뿐만 아니라 경력, 정책, 자금 조달 또는 기술 등 하나의 행동 경로를 식별하십시오.
계속 탐색하세요
Free newsletter
Get the daily AI briefing
Three verified AI stories every weekday morning, written in plain English. Free forever, no ads.
One email each weekday. Unsubscribe in one click. We never sell or share your address.
Test yourself
Take the Federal Agency AI Use Case Inventories quiz
Instant feedback on every answer, and a shareable certificate with a verifiable ID once you pass a course.
Support free AI education. AI Understanding is a 501(c)(3) nonprofit — no ads, no paywall, ever. Make a donation
자주 묻는 질문
What is Federal Agency AI Use Case Inventories?
Federal AI use case inventories make public many ways executive agencies use or plan to use AI, but their reporting rules come from multiple authorities. As of the 2025 cycle, OMB M-25-21 directs most agencies to submit at least annual inventories and publish a public version, while law, security limits and reporting criteria determine what is included.
What does OMB Memorandum M-25-21 direct most agencies to do with AI use cases?
M-25-21 calls for at-least-annual inventory submissions to OMB and public versions on agency websites, subject to stated exceptions.
Which two entities are excluded from M-25-21’s annual inventory task?
The memo explicitly excludes the Department of Defense and Intelligence Community from that inventory requirement.
What kind of information does the Advancing American AI Act say an inventory covers?
Section 7225 describes inventories of agency AI use cases, including current and planned uses.
How should readers interpret OMB’s consolidated repository?
OMB says it compiled the central repository from agency-published inventory information.
Why might an AI use case not appear as an individual public record?
Instructions identify some out-of-scope cases and cases with limits on public release.
계속 학습하세요
관련 가이드
이 주제에 대해 선택된 추가 가이드