Trump Issues National-Security AI Directive, Rescinds Biden-Era Guidance

President Donald Trump on Friday issued a new national security directive ordering the U.S. military and intelligence system to move faster in adopting advanced artificial intelligence, setting a series of 90- and 120-day deadlines for changes to weapons policy, procurement, governance, computing capacity and workforce planning. The unclassified memorandum, dated June 5 and titled “National Security Presidential Memorandum/NSPM-11,” explicitly rescinds and replaces the Biden administration’s 2024 national-security AI directive, NSM-25.

The shift matters because it reaches across the core machinery of the national security enterprise: how agencies buy leading AI systems, how they update rules for autonomous weapons, how they expand access to advanced computing power, how they share data and models across classified environments, and how they recruit and train technical talent. Many of the required steps are subject to the availability of appropriations.

The memorandum, signed by Trump and titled “Artificial Intelligence in the National Security Enterprise,” says AI will be central to future U.S. power. “Artificial intelligence (AI) will be among the most transformative technologies to national security in the history of the United States,” the White House said in Section 1 of the memo.

NSPM-11 organizes policy around four pillars: Adoption, Adaptation, Assurance and Accountability. In broad terms, the order pushes agencies to deploy advanced AI more quickly, adjust institutions and processes around it, build confidence in how systems are tested and used, and impose policy limits on government use.

Under the accountability section, the memo draws a line on civil-liberties and viewpoint concerns. “American AI technologies shall neither be developed nor used by the national security enterprise to censor free speech, embed ideological bias, or conduct unauthorized or unlawful surveillance activities,” the White House said in Section 2(d).

Among the first deadlines, the memo directs the Secretary of War — the administration’s current term for the Pentagon chief — to update the Defense Department’s Directive 3000.09 on autonomy in weapon systems within 90 days and review it annually after that. In the same 90-day window, the Committee on National Security Systems and the Office of Management and Budget, working with the White House Office of Science and Technology Policy, must issue a policy for governing AI use in national-security systems.

The memorandum also orders a classified annex within 90 days to address sensitive issues not covered in the public document. That means part of the administration’s implementation plan will remain nonpublic even as the White House lays out its broader direction in unclassified form.

Within 120 days, the Secretary of War, the director of national intelligence and other agencies with intelligence components are directed to review and update procurement processes to speed the onboarding of the most advanced AI models from multiple vendors. The memo also says agencies should, to the maximum extent permitted by law, terminate contracts for default or convenience with companies that have “repeatedly demonstrated a pattern of conduct” inconsistent with the memo’s policies, with waivers capped at no more than one year.

On computing capacity, the White House ordered officials within 90 days to produce a roadmap for ensuring national-security agencies have adequate access to advanced computing resources. That includes possible advanced AI computing facilities and an AI test range, subject to appropriations.

The memo also pushes more sharing across the defense and intelligence system. Within 120 days, the director of national intelligence, the Secretary of War, the Energy secretary and the director of the National Security Agency must begin joint AI data and model exchanges that can be accessed across multiple enclaves for shared mission uses.

Workforce and training are another major piece of the order. Agencies are told to speed the hiring of technical AI talent, while the Office of Personnel Management must begin creating an “AI National Security Strategic Reserve” of nongovernmental AI talent within 120 days. The memo also calls for an “AI for National Security Curriculum” within 120 days, along with joint AI risk-management measures and standardized test, evaluation, verification and validation methods, often referred to as TEVV.

The policy marks a formal break from the Biden administration’s national-security AI framework. NSM-25, issued in October 2024, emphasized “safe, secure, and trustworthy” AI development, human-rights protections and independent review for national-security uses. NSPM-11 says it rescinds and replaces that memo and its associated guidance.

It also comes three days after a separate White House executive order, dated June 2, called “Promoting Advanced Artificial Intelligence Innovation and Security,” which pushed broader federal AI security and public-private cooperation measures, including a voluntary framework for federal access to certain frontier AI models before public release. With Friday’s memorandum, the administration has now paired that broader AI policy push with a detailed national-security directive, although some implementation details will remain classified and some actions will depend on funding approved by Congress.