Below is an updated overview of the federal government’s most consequential artificial-intelligence guidance documents, current as of August 2025. These policies collectively accelerate AI innovation while preserving accountability and public trust.
OMB Memorandum M-25-21 — Accelerating Federal Use of AI through Innovation, Governance, and Public Trust
Issued: April 3 2025 | Authority: Office of Management and Budget | Scope: All Executive-Branch agencies (including independent regulatory agencies)
Core Objective: Maintain U.S. global AI leadership while safeguarding civil rights, civil liberties, and privacy. The memorandum mandates a forward-leaning, pro-innovation posture and slashes bureaucratic hurdles that slow deployment.
| Deadline | Agency Action Required | Applies To |
|---|---|---|
| 60 days (June 2 2025) | Designate a Chief AI Officer (CAIO) | All agencies |
| 90 days (July 2 2025) | Establish an AI Governance Board | CFO Act agencies |
| 90 days (July 2 2025) | Launch the inter-agency Chief AI Officer Council | Chaired by OMB |
| 180 days (Sept 30 2025) | Publish a public-facing AI Strategy | CFO Act agencies |
| 180 days (Sept 30 2025) | Submit compliance implementation plans | All agencies |
| 270 days (Dec 29 2025) | Update IT, data, and cybersecurity directives to align with AI use | All agencies |
| 270 days (Dec 29 2025) | Issue a generative-AI policy | All agencies |
| 365 days (Apr 3 2026) | Fully implement high-impact AI risk-management regime | All agencies (except IC) |
High-Impact AI refers to systems whose outputs serve as a principal basis for decisions that are legally or materially binding on rights, benefits, or safety. Such systems must undergo strict testing, independent impact assessments, continuous monitoring, and provide human appeals procedures.
EOIR Policy Memorandum PM 25-40 — Use of Generative Artificial Intelligence in EOIR Proceedings
Issued: August 8 2025 | Authority: Executive Office for Immigration Review (Acting Director Sirce E. Owen) | Scope: All EOIR courts and practitioners
Current Posture: No blanket ban and no mandatory disclosure of AI usage. However, judges may impose local rules, and practitioners face discipline if they file hallucinated citations or other inaccurate, AI-generated materials.
| Risk / Obligation | Key Details |
|---|---|
| Primary Risk | Hallucinated citations or arguments may lead to sanctions and reputational damage. |
| Professional Duty | Verify every AI-generated citation; comply with state-bar ethics rules; consult clients when appropriate. |
| Adjudicator Role | Remain vigilant; report suspected misconduct to the Attorney Discipline and Anti-Fraud programs. |
| Future Guidance | DOJ-wide generative-AI policy expected by December 29 2025; BIA may issue precedential decisions. |
Context & Legal Framework
Both memoranda align with Executive Order 14179 — “Removing Barriers to American Leadership in Artificial Intelligence” (January 23 2025), which rescinded the previous administration’s AI order and set a policy of rapid, responsible AI adoption.
Chief Justice John Roberts echoed this balance of innovation and caution in the 2023 Year-End Report on the Federal Judiciary, noting that AI “has great potential to dramatically increase access to key information” but “requires caution and humility.”
Verified Source Documents
| Document | Date Issued | Official PDF / Web Link |
|---|---|---|
| OMB Memorandum M-25-21 | April 3 2025 | Download PDF |
| EOIR Policy Memorandum PM 25-40 | August 8 2025 | Download PDF |
| Executive Order 14179 — Removing Barriers to American Leadership in AI | January 23 2025 | View EO |
| Supreme Court 2023 Year-End Report on the Federal Judiciary | December 31 2023 | Download PDF |
