Trump's NSPM-11 Reshapes How America's Military and Intelligence Agencies Use AI
President Trump signed National Security Presidential Memorandum-11 on June 5, 2026, directing the US military and intelligence community to accelerate AI adoption while setting civil liberties limits and moving to establish an AI National Security Strategic Reserve. The directive rescinds and replaces the Biden-era NSM-25.
A Directive Designed for Speed
On June 5, 2026, President Donald Trump signed National Security Presidential Memorandum-11, titled “Artificial Intelligence in the National Security Enterprise.” The memo rescinds and replaces Biden’s NSM-25.
The core thesis of the document is acceleration. The memo says previous administrations imposed undue bureaucracy that hampered the pace of AI adoption, and that competitors have continued to develop and deploy AI for military and intelligence purposes.
“The United States can and will responsibly accelerate the use of AI across intelligence and warfighting domains in line with American values,” the memorandum states. That framing — acceleration as responsibility, not risk — runs through every substantive provision in the document.
Four Pillars
NSPM-11 organizes its mandates around four principles the administration calls Adoption, Adaptation, Assurance, and Accountability.
Adoption directs national security agencies to eliminate institutional barriers to deploying advanced AI models. Agencies are instructed to maintain deep, proactive partnerships with industry to make the most advanced frontier models broadly available to national security professionals without delay. The memorandum gives the Secretary of War, the Director of National Intelligence and the heads of agencies with IC elements 120 days to update procurement processes to enable rapid onboarding of “the most advanced AI models from multiple vendors.”
Adaptation directs the national security enterprise to adapt commercial or open-source AI technologies from diverse suppliers. Agencies are directed to adapt commercial or open-source AI technologies, and where a commercial solution is not appropriate for security or mission reasons, they may deploy customized technologies or develop AI internally. This is a recognition that the diversity of AI capabilities in 2026 — from reasoning models to multimodal systems to code generation — requires mission-specific tool selection.
Assurance addresses the reliability and security requirements unique to national security AI use. Systems must be “reliable, robust, steerable, and controllable.” Critically, the memo requires, through contractual clauses or other means, that no commercial entity or adversary can prevent use of, disable, degrade, or materially modify an AI system that warfighters depend on without Federal Government knowledge and approval.
Accountability is the civil liberties anchor of the directive. The memo explicitly prohibits AI deployed in national security contexts from being used to censor free speech, embed ideological bias, or conduct unauthorized surveillance of American citizens. These prohibitions are framed as hard limits, not policy preferences — a deliberate move to preempt concerns that AI-enabled government surveillance would expand under the new directive.
The AI National Security Strategic Reserve
One of the more unusual provisions of NSPM-11 is the effort to establish an “AI National Security Strategic Reserve” — a reserve of non-governmental AI talent to support federal efforts on AI national security issues “as needed.”
The memo gives the OPM Director 120 days to initiate efforts to establish the Reserve.
This signals a recognition that the federal government’s internal AI talent pipeline remains insufficient relative to the scale of what NSPM-11 envisions. The Strategic Reserve is an attempt to bridge that gap without requiring full-time federal employment.
What NSPM-11 Replaces — and Why It Matters
Biden’s NSM-25, the memorandum NSPM-11 supersedes, directed the national security enterprise to adopt AI while protecting human rights, civil rights, civil liberties, privacy, and safety.
NSPM-11 doesn’t eliminate governance — the Accountability pillar preserves core civil liberties protections — but it shifts the default posture. Where NSM-25 said “prove it’s safe before deploying,” NSPM-11 effectively says “deploy and govern simultaneously.” Many provisions carry 90- or 120-day implementation deadlines.
For intelligence agencies, this matters in concrete operational terms. The memo calls for rapid onboarding of AI models from multiple vendors and criticizes previous administrations for fostering dangerous dependencies on single vendors.
The Geopolitical Stakes
NSPM-11 lands in a context where AI capability has become explicitly geopolitical.
By directing rapid procurement, multi-vendor diversification, a talent reserve, and simultaneous development of governance guardrails, the administration is attempting to compress the timeline between “AI capability exists commercially” and “AI capability is operational in US national security systems.”
Prohibiting AI from being used for unauthorized surveillance is meaningless if there is no independent body with authority and resources to verify compliance.
Industry Implications
For AI companies with government contracts or aspiring to them, NSPM-11 dramatically expands the addressable market for frontier model deployment in national security applications. The directive’s emphasis on multi-vendor procurement and rapid onboarding — combined with its explicit push to get state-of-the-art commercial models into classified environments — signals that the government AI market is about to become significantly more competitive and significantly larger.
The memo’s “multi-vendor” language is also notable for what it implies about risk concentration. A government that depends entirely on a single AI provider for critical intelligence functions faces an unacceptable dependency. By mandating vendor diversity, NSPM-11 creates structural demand for multiple frontier AI providers to maintain government-facing offerings — a dynamic that benefits the broader frontier AI ecosystem rather than concentrating advantage in any single company.
What NSPM-11 leaves unanswered is perhaps as important as what it directs. The balance between speed and safety in AI deployment for warfighting applications is not a question that 120-day procurement timelines resolve. Getting that balance wrong — in either direction — carries consequences that extend well beyond any single government contract.
Correction (November 15, 2026): 24 sentences were corrected or removed to match the sources listed on this page.