AI Regulation Tracker / Executive action
NSPM-11: Trump AI Memo for the National Security Enterprise
According to a White House fact sheet, President Trump signed National Security Presidential Memorandum-11 on June 5, 2026, directing the national security enterprise to accelerate AI adoption, ordering an updated directive on autonomy in weapon systems, and barring any vendor from disabling, degrading, or modifying an AI system that American warfighters depend on without prior approval. The fact sheet says the memorandum rescinds and replaces the Biden Administration's NSM-25. The underlying NSPM-11 text is not public; the fact sheet describes the directive as signed and effective as of June 5, 2026.
The White House released NSPM-11 on June 5, 2026 under the title "Artificial Intelligence in the National Security Enterprise." According to the official fact sheet, the memorandum directs the national security enterprise to accelerate AI adoption to meet surging demand, to adapt the best commercial and open-source technologies for mission use, and to keep fielded systems secure. It also calls for building out secure computing facilities to run future AI systems at scale and for strengthening the talent pipeline, including an AI National Security Strategic Reserve of top non-governmental experts.
Two provisions carry the most weight for anyone doing business with the defense or intelligence buyer. The first is a vendor-control rule. The second is a rewrite of how the Pentagon governs autonomy in weapons. Both are worth reading closely.
What does the vendor-control provision actually say?
The fact sheet states that the memorandum "directs departments and agencies to ensure that no entity, commercial or otherwise, can disable, degrade, or modify an AI system that American warfighters depend on without prior approval." Read that as written. It is aimed at the scenario where a private company can reach into a deployed system and turn it off, throttle it, or change how it behaves, and it says that cannot happen without the government's sign-off first.
The stated reason for it sits in the same document. The White House describes the prior Biden memorandum, NSM-25, as having "fostered dangerous single-vendor dependencies that left our warfighters exposed." So the concern being addressed is lock-in: a mission that depends on one supplier's system, where that supplier retains the practical ability to disable or alter it. My read is that this pushes real contractual and architectural obligations onto vendors around update control, remote administration, and kill-switch design. If your product can be changed from your side after it is fielded, expect the government to want a say in when and whether that happens.
What changes for weapons autonomy?
The memorandum "directs the Secretary of War to issue an updated directive on autonomy in weapon systems and requires annual review of key guidance across the national security enterprise." This is the part to watch, and also the part not to overstate. The updated autonomy directive is ordered, not written. As of this writing it has not been issued, so nobody outside the process knows its terms. What the memorandum does is set that rewrite in motion and put the enterprise on an annual review cadence for its key guidance. The direction of travel is faster adoption with refreshed rules, but the operative document that will govern autonomy is still to come.
How does this differ from the Biden approach?
NSPM-11 "rescinds and replaces the Biden Administration's NSM-25," which the fact sheet characterizes as an outdated document that "burdened American AI adoption with ideological mandates" and created single-vendor dependencies. The framing is a deliberate reversal. Where the prior posture is described as a brake, this memorandum is pitched as an accelerator, paired with a security and control layer on top. It also offers new partnerships with willing private-sector companies to help secure US AI against foreign threats, which signals that the government wants closer coordination with industry, not distance from it.
Why does this reach beyond the Beltway?
Most professionals will never touch a defense contract. The reason this still matters is that the national security buyer sets tone for the wider market. When the government says a supplier cannot silently disable or alter a fielded AI system that people rely on, and ties that to procurement, it puts a name to a governance question every serious AI buyer is starting to ask. Who controls the off switch? Who can change the model after it is in production? Can the customer see and approve those changes? Those are not defense-only questions. They are the same questions a hospital, a bank, or a law firm should ask a critical AI vendor. This memorandum is the clearest signal yet that the largest US buyer is treating vendor control as a first-order risk, and commercial buyers tend to follow that lead.
What should a vendor or buyer do now?
If you sell AI into national security programs, start by identifying which of your systems could be characterized as something warfighters depend on, then trace every mechanism by which your company could disable, degrade, or modify those systems after deployment. That is the surface the prior-approval rule touches. On the buyer side, whether federal or commercial, the practical lesson is to write control into the contract rather than assume it: who may change or disable the system, under what notice, and with whose approval. AI augments the judgment of the people running these programs. It does not replace their responsibility to know who holds the controls.
Questions professionals are asking
Is NSPM-11 in force now, or is it a plan?
According to the White House fact sheet, President Trump signed it on June 5, 2026 and the fact sheet describes it as effective, directing the national security enterprise. The full NSPM-11 text is not public, so the fact sheet is the operative official summary. One thing it orders, an updated directive on autonomy in weapon systems, is a future deliverable that had not been issued as of this writing, so that specific rule is coming rather than here.
Does the memorandum ban vendors from updating their systems?
It does not ban updates outright. The fact sheet says agencies must ensure that no entity can disable, degrade, or modify an AI system that warfighters depend on without prior approval. The control point is unilateral vendor action on a relied-upon system without the government signing off first.
What happened to the Biden AI national security memorandum?
NSPM-11 rescinds and replaces the Biden Administration's NSM-25. The White House fact sheet describes NSM-25 as an outdated document that slowed AI adoption and fostered single-vendor dependencies. The new memorandum reframes the posture toward faster adoption with added security and control conditions.
Does this affect companies that do not sell to the military?
Not directly as law. Its reach is the federal national security enterprise and its contractors. The wider relevance is that the largest US buyer is now treating control of a fielded AI system as a named procurement risk, which is the same vendor-control question commercial buyers in regulated sectors increasingly ask.
RELATED BRIEFINGS
Browse the full AI Regulation News tracker
Informational analysis for working professionals, not legal advice. Confirm how any directive, statute, or requirement applies to your situation with qualified professionals in the relevant jurisdiction.