On June 5, 2026, President Trump issued a national security presidential memorandum (NSPM-11) directing national security agencies to rapidly adopt frontier AI. Coming just three days after the signing of an executive order on cybersecurity evaluation of frontier AI models, it is framed as a direct response to the contract dispute between Anthropic and the Department of Defense (DoD).
June 5, 2026 · National Security Presidential Memorandum
NSPM-11: The Pentagon Moves to Lock In Its AI
A new Trump-administration directive orders rapid military adoption of AI — and bars commercial vendors or adversaries from disabling, degrading, or altering the systems warfighters depend on without federal approval. It rescinds and replaces the Biden-era NSM-25.
NSPM-11
Replaces NSM-25 (Oct 2024)
90 days
Directive 3000.09 update + classified annex (~Sept 2026)
3 days
After the June 2 Executive Order on frontier AI safety
Multi
Multiple vendors to avoid single points of failure
The core rule
Through contract clauses, no company or adversary may disable, degrade, or substantially alter a military-relied AI system — without federal approval.
Rapid deployment of commercial AI on classified networks
Update of autonomy directive so AI respects chain of command
Creation of a new AI talent reserve
The backdrop · Anthropic vs. Department of War
Talks collapse
Anthropic holds two "red lines": no mass surveillance, no fully autonomous weapons
→
Late Feb 2026
Designated a supply-chain risk; orders to halt dealings with agencies
→
March 2026
Anthropic sues (N.D. California and other venues)
SUPPORTERS
Rapid military AI adoption strengthens defense and secures U.S. advantage; multi-vendor sourcing reduces dependence on any single firm.
CRITICS
Human rights groups warn the rule could override corporate "red lines" and expand autonomous weapons; OpenAI and Google have also been cautious on military use.
The central question: Can AI companies impose their own ethical standards on government use? Anthropic calls the designation a "First Amendment violation"; the DoW calls it a "national security risk." As a policy memo, NSPM-11 sets no benchmarks or prices — the fight over "control" looks set to continue.
Continue reading The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in ✓ Signed in — this article isn’t included in your current plan.Unlocking the full article…