MEMORANDUM FOR THE SECRETARY OF STATE
THE SECRETARY OF THE TREASURY
THE SECRETARY OF WAR
THE ATTORNEY GENERAL
THE SECRETARY OF ENERGY
THE SECRETARY OF HOMELAND SECURITY
THE DIRECTOR OF THE OFFICE OF MANAGEMENT AND BUDGET
THE DIRECTOR OF NATIONAL INTELLIGENCE
THE DIRECTOR OF THE CENTRAL INTELLIGENCE AGENCY
THE ASSISTANT TO THE PRESIDENT FOR NATIONAL SECURITY AFFAIRS
THE ASSISTANT TO THE PRESIDENT FOR SCIENCE AND TECHNOLOGY
THE ASSISTANT TO THE PRESIDENT FOR POLICY AND HOMELAND SECURITY ADVISOR
THE DIRECTOR OF THE FEDERAL BUREAU OF INVESTIGATION
THE DIRECTOR OF THE OFFICE OF PERSONNEL MANAGEMENT
THE NATIONAL CYBER DIRECTOR
SUBJECT: Artificial Intelligence in the National Security Enterprise
By the authority vested in me as President by the Constitution and the laws of the United States of America, I hereby direct the following:
Section 1. Purpose.
Artificial intelligence (AI) is set to be one of the most impactful technologies in the history of U.S. national security. When implemented correctly, AI can safeguard our military personnel during both peace and conflict, facilitate precise operations to minimize civilian casualties, and ensure the U.S. maintains a technological advantage over adversaries and strategic competitors.
Previous administrations have created excessive bureaucratic hurdles that slowed AI adoption, led to over-reliance on individual vendors, and obstructed the military’s access to cutting-edge technologies. Meanwhile, competitors have advanced their own AI and autonomous technologies for military and intelligence use, often without adequate oversight or concern for civil liberties.
My Administration is committed to responsibly advancing AI use within intelligence and military operations, aligning with American values. The U.S. military is renowned for its effectiveness and moral standards, earning trust through a consistent chain of command and accountability, from government leadership to the operatives executing missions.
We will ensure our national defenders have access to advanced and secure AI technologies for critical missions. This includes streamlining the acquisition and deployment processes, maintaining rigorous oversight, and ensuring a robust supply chain that withstands conflict. Collaboration with private sectors and academia will enhance our technological expertise, ensuring our forces are well-trained in utilizing AI systems. These efforts aim to secure a lasting AI advantage while upholding constitutional governance.
Sec. 2. Policy.
My Administration will drive the development and deployment of AI for national security, focusing on four core areas:
(a) Adoption. The national security enterprise will expedite AI adoption by pinpointing areas where AI can boost operational efficiency and removing unnecessary barriers. This includes fostering deep partnerships with industry to provide cutting-edge models to national security professionals promptly, ensuring technological superiority and encouraging rapid experimentation and validation.
(b) Adaptation. The national security enterprise will adapt commercial or open-source AI technologies, utilizing the most advanced capabilities from diverse suppliers, while ensuring these technologies are optimized for their intended purposes. Where commercial solutions are unsuitable, agencies may deploy or develop customized AI technologies. These technologies will be made available across the national security enterprise to support various missions.
(c) Assurance. The enterprise will ensure all AI technologies are reliable, robust, steerable, and controllable, operating under applicable laws and policies. Measures will be taken to prevent any commercial entity or adversary from interfering with AI systems crucial to mission success. Comprehensive security and functionality measures will ensure AI systems’ confidentiality, integrity, reliability, availability, and interoperability.
(d) Accountability. AI technologies will not be used to censor free speech, establish ideological bias, or conduct unauthorized surveillance. AI use must align with U.S. civil liberties and privacy protections. Commanders and agency leaders will be accountable for upholding these standards as AI capabilities and privacy regulations evolve.
Sec. 3. Updated Policies and Guidance.
(a) Within 90 days, the Secretary of War will update DOD Directive 3000.09 on Autonomy in Weapon Systems, ensuring AI systems respect the chain of command and operational authorities, consistent with this memorandum’s policies.
(b) Per the Federal Information Security Modernization Act of 2014, the Secretary of War, the DNI, and relevant agency heads will terminate contracts with companies that consistently violate policies. Exceptions may be granted for essential operations, with all exceptions reported to the Assistant to the President for Science and Technology and the Assistant to the President for National Security Affairs within 30 days.
(c) Within 90 days, the Committee on National Security Systems and the OMB Director, in coordination with the APST, will issue governance policies for AI use in national security systems, aligning with non-national security system requirements where appropriate.
(d) A classified annex addressing sensitive national security issues will be issued within 90 days.
(e) Agency heads will update policies to align with this memorandum, reviewing and updating them annually to reflect AI technology advancements.
(f) This memorandum rescinds and replaces National Security Memorandum-25 and related guidance.
Sec. 4. Advancing National Security Capabilities.
(a) Within 120 days, the Secretary of War, the DNI, and relevant agency heads will update procurement processes for rapid onboarding of advanced AI models from multiple vendors, bridging the gap between public and national security access.
(b) Within 90 days, the APST and the OMB Director, in coordination with relevant agencies, will develop a roadmap for access to advanced computing resources, commissioning AI facilities with high-security requirements and establishing an AI test range for national security use.
(c) Within 120 days, the Secretary of War, the Secretary of Energy, the DNI, and the NSA Director will develop partnerships with private-sector companies to secure advanced AI technologies, including collaborations on security R&D and threat intelligence sharing.
(d) The Secretary of Energy will collaborate with relevant agencies through the Genesis Mission to enhance AI capabilities for national security missions.
(e) The DNI, in coordination with IC elements, will prioritize collecting and analyzing foreign AI technologies and develop strategies with the Secretary of State to engage allies and partners.
(f) Within 120 days, the DNI, the Secretary of War, the Secretary of Energy, and the NSA Director will initiate joint AI data and model exchanges for national security applications.
Sec. 5. Building Capacity for AI Adoption.
(a) Agencies will utilize special hiring and pay authorities to recruit technical AI talent into the Federal Government.
(b) Within 120 days, the OPM Director, in coordination with relevant agencies, will establish an AI National Security Strategic Reserve of non-governmental AI talent to support federal AI national security efforts.
(c) Within 120 days, the DNI and the Secretary of War will develop an AI for National Security Curriculum, ensuring personnel are trained to use AI systems and understand their implications for national security.
(d) Agencies will prioritize R&D for AI reliability, robustness, steerability, and controllability, maintaining leadership in this domain.
(e) Within 120 days, a joint strategy for AI risk management will be developed to establish baseline security practices for the national security enterprise.
(f) Within 120 days, standardized AI national security test, evaluation, verification, and validation methodologies will be submitted for review.
Sec. 6. Definitions.
(a) “Artificial intelligence” or “AI” is defined as per 15 U.S.C. 9401(3);
(b) “AI incident response” refers to handling performance degradation or data loss in AI systems;
(c) “AI security” involves protection measures across the AI technology stack;
(d) “AI technology stack” includes hardware, infrastructure, AI models, and applications;
(e) “Chain of command” refers to the succession of authority in achieving lawful objectives;
(f) “Controllability” is the ability to monitor and adjust system operations as needed;
(g) “Intelligence Community” is as defined in section 3003 of title 50, United States Code;
(h) “National security enterprise” includes the Department of War and other agencies with national security roles;
(i) “Reliability” is a system’s ability to perform without failure under specified conditions;
(j) “Robustness” is a system’s ability to maintain performance under various conditions;
(k) “Steerability” is the ability to influence a system’s internal behavior to meet objectives.
Sec. 7. General Provisions.
(a) This memorandum shall not interfere with the legal authority of any executive department or agency.
(b) Implementation shall comply with applicable law and appropriations.
(c) This order does not create any enforceable rights or benefits against the United States or its entities.
DONALD J. TRUMP
