"The United States can and will responsibly accelerate the use of AI across intelligence and warfighting domains in line with American values," the National Security Presidential Memorandum (NSPM) 11 says.
What NSPM-11 directs: speed, procurement, and limits
Issued Friday, NSPM-11 is a six-page policy edict that directs the federal government to accelerate adoption of AI in national security by streamlining procurement, expanding training, and enhancing security in close collaboration with leading technology firms. The memorandum explicitly calls for maintaining "rigorous oversight … including testing, evaluation, validation, and verification" even as it urges faster deployment.
At the same time, the memo revokes Biden-era guardrails the administration considers too restrictive and orders the "termination for default or for convenience" of contracts with any company that attempts to restrict how the government uses its AI—while allowing strictly limited waivers for pressing needs.
Anthropic, Claude Mythos, and the NSA waiver
Although NSPM-11 avoids naming companies, the memorandum and the administration's recent actions are widely read as aimed at Anthropic. The story in February, as reported, involved Anthropic questioning the reported use of its model to plan combat operations against Venezuela and Iran; after that dispute, the administration began switching to other companies' AIs and moved to cancel Anthropic's federal contracts, citing a "supply chain risk."
Anthropic's Claude was the first commercial large language model authorized for use on classified networks. The memo allows for limited waivers — a complexity underscored by the fact that Anthropic's Claude Mythos remains in use at the NSA under such an exception, showing the administration's simultaneous push for access and leverage.
Legal dueling lines: lawsuits, contract language, and a possible "do‑over"
Anthropic challenged the contract cancellations in two parallel lawsuits. Experts who spoke to Breaking Defense are divided over how NSPM-11 affects those cases. Jack Shanahan, who founded Project Maven and the Pentagon’s Joint AI Center, said the memo "comes too late to affect a judge’s ruling" on what was true at the time the suits were filed, but it could give the administration a "do‑over"—allowing it to try cancelling contracts again under the new policy if it loses in court.
By contrast, Jessica Tillipman, associate dean at George Washington University and a federal procurement lawyer, said she did not read the memo as a litigation hedge. Instead, Tillipman argued the memorandum is addressing broader problems the government experienced with Anthropic, especially "dependency on single suppliers," and that lasting change will require implementing new language "contract by contract." She warned that the specific wording of clauses that follow will determine whether the government can more easily remove a contractor, including subcontractors.
Former CIA CTO Nand Mulchandani framed the Anthropic dispute as a contracting problem "that got out of hand" and said NSPM-11 is essentially guidance for officials negotiating future contracts: "it depends on the contract you write."
DoDD 3000.09: a 90‑day rewrite and the mechanics of "responsible" AI
NSPM-11 tasks Defense Secretary Pete Hegseth with submitting a revised Defense Department directive, DoDD 3000.09, "Autonomy in Weapons Systems," within 90 days. The current version was last updated under the prior administration in 2023. Shanahan called the deadline tight but achievable if revisions are modest; he warned a sweeping overhaul completed in three months would be difficult to do properly.
Observers quoted in the reporting stressed continuity with earlier policy despite the administration's rhetoric. Mike Horowitz, a former Pentagon policy official who helped author the Biden-era policies, said that beyond the rhetoric there is "a lot more continuity here than change," and that NSPM-11 still emphasizes testing, steerability, and controllability.
How technologists, procurement leaders, and the Pentagon are likely to respond
- Technologists and security teams will have to maintain rigorous testing and validation processes as NSPM-11 stresses technical evaluation even while accelerating deployments.
- Procurement leaders and contract negotiators will focus on clause language: Tillipman and Mulchandani both emphasized that the memorandum's impact will hinge on the contracts written afterward and whether they enable easier termination of suppliers and subcontractors.
- The Pentagon and Defense leadership will be watching the 90‑day rewrite of DoDD 3000.09 closely; Shanahan noted companies will "breathe a sigh of relief" if the revision is reasonable and modest, but warned that "speed without proper test and evaluation is a catastrophe waiting to happen."
NSPM-11 is both a signal and a tool: it pushes agencies to move faster with AI while carving out sharper leverage over suppliers and insisting on oversight. The immediate legal fights over Anthropic's canceled contracts will play out in court; whether NSPM-11 ultimately reshapes those outcomes depends on the contract language that follows and on whether the Defense Department's expedited rewrite of DoDD 3000.09 is narrow or sweeping.
