Trump orders faster military AI and bars vendors from switching off fielded models
A new national security memo tells the Pentagon to onboard frontier commercial AI at speed, strips vendors of the power to shut it down mid-mission, and resets the rules for autonomous weapons within 90 days.
A new national security memo tells the Pentagon to onboard frontier commercial AI at speed, strips vendors of the power to shut it down mid-mission, and resets the rules for autonomous weapons within 90 days.
President Donald Trump signed a national security memorandum, NSPM-11, on Friday directing the military and intelligence agencies to accelerate their use of frontier artificial intelligence, the White House said.
The sharpest line is a control clause. No commercial entity or adversary may disable, degrade, or materially modify an AI system that warfighters depend on without federal knowledge and approval, the memo states, language the administration frames as a supply chain that cannot be severed in wartime. The Next Web called that vendor restriction the most striking provision: a company can no longer pull a deployed model from military use on its own, even over safety concerns.
That reads as a direct answer to the Pentagon's fight with Anthropic. The company wanted contract assurances that the military would not use its Claude model in fully autonomous weapons or to surveil Americans. Defense Secretary Pete Hegseth said it had to allow any use the Pentagon deemed lawful, moved to label Anthropic a supply chain risk, and Anthropic sued, the Washington Times reported.
NSPM-11 pushes the other way. It orders rapid onboarding of the most advanced models from multiple vendors, after classified Pentagon deals with Nvidia, Microsoft, and AWS broke Anthropic's early grip on classified use.
The battlefield stakes are concrete. The US military ran Claude during the January operation to capture Venezuelan leader Nicolás Maduro, the first known use of an Anthropic model in a classified mission, through an Anthropic-Palantir tie-up, Indian Express reported, citing the Wall Street Journal. AI is compressing the time to find and strike a target.
The battlefield and the startup story — free in your inbox every week. No paywall.
Subscribe Free →Hegseth now has 90 days to rewrite Defense Department Directive 3000.09, the rule governing autonomy in weapons and the human judgment required before lethal force. The memo says any update must respect the chain of command.
Frequently Asked Questions
What is NSPM-11?
It is a national security presidential memorandum Trump signed on Friday, June 5, directing the military and intelligence agencies to accelerate their use of frontier artificial intelligence, according to the White House and The Next Web. It rescinds and replaces the earlier National Security Memorandum-25.
What is the memo's most contested provision?
A control clause stating that no commercial entity or adversary may disable, degrade, or materially modify a fielded AI system without federal knowledge and approval. The Next Web called it the most striking provision, because a vendor can no longer pull a deployed model from military use on its own, even over safety concerns.
How does this connect to Anthropic?
The Washington Times reported that Anthropic wanted contract assurances its Claude model would not be used in fully autonomous weapons or to surveil Americans. Defense Secretary Pete Hegseth said the company had to allow any use the Pentagon deemed lawful, moved to label it a supply chain risk, and Anthropic sued.
What is Defense Department Directive 3000.09?
It is the Pentagon policy governing autonomy in weapon systems, including the human judgment required before lethal force. The memo gives Hegseth 90 days to issue an updated directive that, per the Washington Times, must respect the chain of command.
Has the US military already used commercial AI in an operation?
Indian Express, citing the Wall Street Journal, reported that the US military ran Anthropic's Claude during the January operation to capture Venezuelan leader Nicolás Maduro, the first known use of an Anthropic model in a classified mission, through an Anthropic-Palantir partnership.
AI-generated summary, reviewed by an editor. More on our AI guidelines.
Marcus Schuler
Marcus Schuler edits BattlePolicy, a daily defense-technology brief connecting the companies and capabilities behind modern war to the contest among Europe, the US, Russia, and China.