National Security Presidential Memorandum 11: Artificial Intelligence in the National Security Enterprise

Record US-28 · NSPM-11

Issued by: The White House (President Donald J. Trump)
Jurisdiction: United States
Level · region: National · North America
Published / adopted: 2026-06-05
In force: 2026-06-05
Language: English
Official identifier: NSPM-11

Summary

Presidential memorandum that rescinds and replaces NSM-25 and sets AI policy for the national security enterprise under four pillars: adoption, adaptation of commercial and open-source models, assurance (AI that is reliable, robust, steerable and controllable, that no vendor or adversary can disable or alter without government approval, and that is tested and evaluated) and accountability (no AI for censorship, ideological bias or unlawful surveillance). It orders an update of DoD Directive 3000.09 on autonomy in weapon systems, termination of contracts with companies repeatedly acting against this policy, a joint AI risk-management strategy and AI test and evaluation methods.

Safety topics

national security & militaryevaluation & testinghuman oversightaccountability & liabilitygeneral-purpose / frontier AI

Status notes

Section 3(f) rescinds and replaces NSM-25 (US-09) and associated guidance. The text does not use the word 'safety'; its safeguards are framed as reliability, robustness, steerability, controllability, security practices and civil liberties, terms it defines in Sec. 6. Deadlines: within 90 days (about 3 Sep 2026), the updated DoD Directive 3000.09, a policy of the Committee on National Security Systems and OMB for governing AI use in national security systems, a roadmap for advanced computing and an AI test range, and a classified annex; within 120 days (about 3 Oct 2026), updated procurement processes, private-sector partnerships to secure frontier AI (including joint red-team exercises), joint AI data and model exchanges, an AI National Security Strategic Reserve of non-governmental talent, an AI for National Security Curriculum, the joint AI risk-management and assurance strategy and the test, evaluation, verification and validation methodologies. Agencies may grant waivers of up to one year from the contract terminations under Sec. 3(b).

Sources

Open the official source https://www.whitehouse.gov/presidential-actions/2026/06/national-security-presidential-memorandum-nspm-11/

Secondary source: https://www.whitehouse.gov/fact-sheets/2026/06/fact-sheet-president-donald-j-trump-signs-historic-directive-on-ai-in-the-national-security-enterprise/

Local copy

Archived. Last attempt 2026-10-17 22:12:24 UTC

Official source (third-party copy, not published) archive/US-28/2026-10-17/official.html

Retrieved 2026-10-17 21:40:48 UTC

SHA-256 db087469eb15cb3dbd2ae5e1a382b13555aa09b2a99b4a36b6c37c09290dc5ad

Secondary source (third-party copy, not published) archive/US-28/2026-10-17/secondary.html

Retrieved 2026-10-17 21:40:51 UTC

SHA-256 553d57edaecc24dfa761b2370157244322ee7bd767afb0021bbd0851f0f3d2b9

Verification

The instrument's own official text, or an official register, was stored and confirms the title, date and status. checked 2026-10-18 · official-url-fetched

Recheck of 17 Oct 2026: re-read the memorandum's text on whitehouse.gov stored on 17 Oct 2026 and the White House fact sheet of 5 Jun 2026 (both differ from the 11 Oct copies only in site navigation). They confirm the title, NSPM-11, the date of 5 Jun 2026, the addressees and Secs. 1 to 7, including the rescission of NSM-25 in Sec. 3(f). issuing_body now follows the form of the register's other presidential instruments. The summary was shortened to the 700-character limit and now includes the contract terminations of Sec. 3(b). The earlier notes said the rescinded guidance included the Framework to Advance AI Governance and Risk Management in National Security and that the annex and updated directive are not public; neither is in the stored sources, so both were removed. The stored sources do not show whether the 90- and 120-day deliverables have been issued.

Second check: Independent check by Claude Code (claude-sonnet-5-5), 2026-10-18: 9 field(s) supported.

History

2026-10-18notes: Section 3(f) rescinds NSM-25 (US-09) and its associated guidance, including the → Section 3(f) rescinds and replaces NSM-25 (US-09) and associated guidance. The t — removed two claims not in the stored sources (that the rescinded guidance included the Framework to Advance AI Governance and Risk Management in National Security; that the annex and updated directive are not public); added the full list of 90- and 120-day deliverables and the Sec. 3(b) waiver2026-10-17-recheck-north-america-1 · Claude Code (claude-opus-5-5)
2026-10-18summary: Presidential memorandum that rescinds and replaces NSM-25 of October 2024 and se → Presidential memorandum that rescinds and replaces NSM-25 and sets AI policy for — the summary was 951 characters, over the 700 limit; shortened, and the Sec. 3(b) contract terminations added, from the stored text2026-10-17-recheck-north-america-1 · Claude Code (claude-opus-5-5)
2026-10-18issuing_body: President of the United States (The White House) → The White House (President Donald J. Trump) — aligned with the register's other presidential instruments of this administration (US-11, US-17, US-20, US-23, US-26, US-29: 'The White House (President Donald J. Trump)'); the memorandum is signed by Donald J. Trump on whitehouse.gov2026-10-17-recheck-north-america-1 · Claude Code (claude-opus-5-5)

Related records

National Security Memorandum on AI (NSM-25) United States · Revoked / withdrawn

America's AI Action Plan United States · Published (non-binding, current)

EO 14179 United States · In force

Citation

The White House (President Donald J. Trump). National Security Presidential Memorandum 11: Artificial Intelligence in the National Security Enterprise [NSPM-11]. 2026-06-05. https://www.whitehouse.gov/presidential-actions/2026/06/national-security-presidential-memorandum-nspm-11/. Status: In force; verified 2026-10-18 (level A). Local copy archive/US-28/2026-10-17/official.html retrieved 2026-10-17 21:40:48 UTC, SHA-256 db087469eb15cb3dbd2ae5e1a382b13555aa09b2a99b4a36b6c37c09290dc5ad. Every AI Policy Ever, version 1.18 (snapshot 2026-10-18), record US-28.

Improve this record

Know of an amendment, a status change or an error? Report news or an error on GitHub; the form opens with this record filled in.