President Donald Trump signed National Security Presidential Memorandum 11 (NSPM‑11) on Friday, marking a decisive push to embed cutting‑edge artificial‑intelligence tools into the U.S. military and intelligence community. The directive orders rapid onboarding of the most advanced AI models from a range of vendors, signaling a shift away from the Pentagon’s historic reliance on a single supplier.
At the heart of the memo lies a provision that bars any AI company from disabling, degrading, or otherwise modifying a model that has been placed into military service without explicit government permission. The clause gives the Department of Defense and its intelligence partners a legal safeguard against sudden loss of capability, even if a vendor raises safety or ethical concerns about how the technology is being used.
The vendor‑restriction provision arrives amid an ongoing dispute with Anthropic, which was recently blacklisted as a supply‑chain risk after refusing to let its Claude models be employed for autonomous weapons or mass‑surveillance tasks. By preventing unilateral pull‑backs, the administration hopes to avoid future interruptions that could compromise operational readiness.
"The men and women who defend our nation deserve the best, most secure and most reliable AI in the world," said Michael Kratsios, director of the White House Office of Science and Technology Policy. His comment underscores the administration’s belief that AI superiority is a national‑security imperative.
In addition to the vendor clause, NSPM‑11 directs Defense Secretary Pete Hegseth to issue an updated directive on autonomous weapon systems within the next 90 days. The forthcoming guidance is expected to amend DoD Directive 3000.09, the cornerstone policy governing the development and deployment of autonomous and semi‑autonomous weapons. The revision will likely reinforce requirements for human judgment before lethal force is applied.
The memorandum also outlines limits on the types of AI applications the government will endorse. Defense agencies are prohibited from creating or deploying models designed to "censor free speech, embed ideological bias or conduct unlawful surveillance against the American people." While the language is broad, the memo does not define those terms or detail enforcement mechanisms, leaving room for interpretation.
NSPM‑11 follows an executive order issued earlier in the week that established a voluntary 30‑day review window for frontier AI models before they are released to the public. Together, the two documents illustrate a dual‑track approach: light‑touch regulation for commercial AI development and an aggressive, government‑driven adoption strategy for national‑security purposes.
The “multiple vendors” language reflects recent classified agreements the Pentagon has signed with Nvidia, Microsoft, and Amazon Web Services, expanding the pool of approved suppliers beyond Anthropic. By formalizing a multi‑vendor framework, the administration aims to mitigate supply‑chain risks and foster competition among AI firms.
Accountability is woven throughout the memo. Commanders, agency heads, and directors are tasked with ensuring AI systems are used in line with the stated obligations. Annual reviews of key guidance across the national‑security enterprise are mandated to keep pace with the rapidly evolving AI frontier. Whether those reviews will be substantive or merely perfunctory remains to be seen.
Critics have warned that the vendor‑restriction clause could limit the government’s ability to address safety concerns promptly, potentially encouraging the deployment of untested or risky systems. Supporters argue that the measure is essential to prevent adversaries from exploiting sudden capability gaps.
As the Pentagon moves to integrate advanced AI models into its arsenal, the balance between rapid innovation and responsible oversight will shape the future of American military technology.
Cet article a été rédigé avec l'assistance de l'IA.
News Factory APP - actualités agentiques pour booster votre SEO et AEO.