Speed Over Caution: What NSPM-11 Means

Signed June 5, 2026, NATIONAL SECURITY PRESIDENTIAL MEMORANDUM/NSPM-11 is the Trump Administration’s most comprehensive statement yet on integrating artificial intelligence into U.S. defense and intelligence operations. Rescinding and replacing the Biden-era NSM-25, the memorandum signals a decisive shift in posture from cautious governance first to speed of adoption first.

Structure

The document is organized around four policy pillars and a cascade of timed directives tasking the Secretary of War, the DNI, NSA, OMB, and OPM with reshaping how AI is procured, deployed, secured, and staffed across the national security enterprise. 

“Previous administrations imposed undue bureaucracy that hampered the pace of AI adoption, fostered dangerous dependencies on single vendors, and made it challenging for our warfighters to adopt the most advanced technologies.”

The memorandum is notable for what it retains from its predecessor (human accountability, civil liberties language) as much as for what it discards (multi-layered pre-deployment review requirements).

We’ve summarized the four pillars for you.

The Four Pillars

Adoption

Eliminate barriers to rapid deployment. Maintain deep industry partnerships. Make frontier models broadly available to national security professionals without delay.

Adaptation

Leverage commercial and open-source AI from diverse suppliers. Customize or internally develop where commercial solutions fall short. Share adapted tools enterprise-wide.

Assurance

Ensure AI is reliable, robust, steerable, and controllable. Contractually prevent vendors from disabling or degrading mission-critical systems without federal approval.

Accountability

AI shall not censor speech, embed ideological bias, or enable unauthorized surveillance. Commanders at every level remain personally accountable for these obligations.

Key Directives — By Timeline

90 days

• Update DOD Directive 3000.09 (Autonomy in Weapon Systems). To be reviewed annually, ensuring autonomous weapons respect the chain of command and operational authorities.
• Contract termination authority. SecWar, DNI, and agency heads directed to terminate contracts with companies demonstrating a “repeated pattern of conduct” inconsistent with NSPM-11 policy. Waivers capped at 1 year; must be reported to APST and APNSA.
• AI governance policy for national security systems. CNSS and OMB Director to issue policy maximizing consistency with OMB M-25-21 where appropriate.
• Classified annex. To be issued addressing sensitive national security dimensions not publicly disclosed.

120 days

• Procurement process review. SecWar, DNI, and IC heads to update acquisition to enable rapid multi-vendor onboarding, closing the capability gap between public and national security access to frontier models.
• Advanced computing roadmap. Assistant to the President for Science and Technology (APST) and OMB Director to develop a roadmap for compute access, including high-security AI computing facilities and an AI test range for national security use cases.
• Private-sector AI security partnerships. SecWar, SecEnergy, DNI, and NSA Director (through AI Security Center) to formalize partnerships for threat intel sharing, joint red-teaming, personnel vetting, and data center security.
• Joint AI data and model exchanges. Cross-enclave data/model sharing for common mission applications across the national security enterprise.
• AI National Security Strategic Reserve. OPM to establish a reserve of non-governmental AI talent available for federal national security efforts.
• AI for National Security Curriculum. DNI and SecWar to develop training ensuring national security personnel understand AI capabilities, limitations, and frontier developments.
• Joint AI risk management strategy. DNI, SecWar, and NSA Director to develop baseline AI security practices for the enterprise, submitted for interagency review before publication.
• Standardized TEVV methodologies. Test, Evaluation, Verification, and Validation standards for high-security AI systems, to be submitted to APST and the National Cyber Director.

Implications & Takeaways

Acquisition

Multi-vendor mandates reshape procurement. Program managers should expect new contracting vehicles designed for rapid AI onboarding.

Autonomy

DODD 3000.09 revision is the pivotal near-term deliverable. The previous directive’s requirement for appropriate levels of human judgment over the use of force has long generated friction between acquisition speed and legal/ethical review. The 90-day update may relax those thresholds for non-lethal or lower-risk autonomy applications while retaining human-in-the-loop requirements for lethal systems. 

Industry

NSA’s AI Security Center is now formally positioned as the center of enterprise AI defense. This follows the pattern of NSA’s Anthropic engineer integration arrangement, where cleared AI specialists are now embedded to support evaluation and hardening of commercial frontier models for national security contexts. NSPM-11 appears to scale that model enterprise-wide.

Workforce

The Strategic Reserve concept is novel and untested. The practical challenges (including clearance timelines, intellectual property conflicts, and competing private-sector compensation) are substantial. OPM will need creative authority, likely including the special pay and hiring provisions referenced in Sec. 5(a), to make this viable.

Civil Liberties

Accountability language is strong but enforcement mechanisms are thin. Section 2(d)’s prohibitions on AI-enabled censorship, ideological bias, and unauthorized surveillance are among the clearest civil liberties commitments in any U.S. AI policy document. However, unlike the assurance and acquisition provisions, no dedicated oversight mechanism or reporting requirement is established to enforce them. The burden falls entirely on commanders and agency heads.

Competition

“Capability gap” language acknowledges a strategic vulnerability. The directive to close the gap between publicly available frontier models and what national security professionals can access is a tacit admission that classification and procurement bureaucracy have left warfighters behind the commercial curve. Cross-enclave model exchanges (Sec. 4(f)) and the advanced computing roadmap (Sec. 4(b)) are structural attempts to close this gap, but will require sustained resourcing and interagency cooperation.

Bottom Line

The central tension here is that NSPM-11 attempts to simultaneously accelerate adoption and enforce accountability. These are goals that have historically traded off against each other. The memorandum’s architecture assumes that speed and safety can be parallel processes rather than sequential gates. Whether the 90–120 day directive cascade can actually produces coherent, interoperable policy outputs will be the measure of the policy’s success in the near term.