Trump signs a memorandum that places 'the most advanced AI' under military control and prohibits suppliers from withdrawing support.

      TL;DR: Trump signed NSPM-11, which mandates swift military AI adoption and prevents vendors from disabling models without authorization. Hegseth must revise the autonomous weapons policy.

      On Friday, President Trump enacted a national security presidential memorandum that instructs the US military and intelligence agencies to hasten their integration of advanced AI technologies. The memorandum, NSPM-11, creates a framework for the "rapid onboarding of the most advanced AI models from various vendors" and prohibits any company from disabling, degrading, or altering an AI system relied upon by military personnel without prior government consent.

      This vendor restriction is particularly significant, as it prevents an AI company from unilaterally withdrawing a deployed model from military applications, even if the firm has concerns about its safety. This provision is particularly relevant in light of the Pentagon's ongoing conflict with Anthropic, which was labeled a supply chain risk after it declined to permit the use of its Claude models for autonomous weaponry or mass surveillance.

      Michael Kratsios, the director of the White House Office of Science and Technology Policy, stated, “The men and women who defend our nation deserve the best, most secure and most reliable AI in the world.”

      Additionally, the memo requires Defence Secretary Pete Hegseth to update the directive regarding autonomous weapon systems within 90 days. This update will modify DoD Directive 3000.09, the primary Pentagon policy governing the use of autonomous and semi-autonomous weapons, including stipulations for human oversight prior to the application of lethal force.

      There are specified limitations within the memorandum, which forbids defense agencies from creating or deploying AI models intended to “censor free speech, embed ideological bias, or conduct unlawful surveillance against the American populace.” However, these terms are not defined, nor is there an explanation of how compliance will be monitored.

      This directive follows a recent executive order that introduced a voluntary 30-day review period for cutting-edge AI models before public dissemination. Collectively, these two documents outline a dual strategy: minimal regulation for the commercial sector and vigorous adoption within the military.

      The mention of "multiple vendors" indicates a departure from reliance on a single supplier. Recently, Anthropic was the sole AI vendor granted approval for classified military applications, but following the Pentagon’s classified agreements with Nvidia, Microsoft, and AWS, the administration is now establishing a multi-vendor strategy.

      The memorandum emphasizes accountability, holding commanders, directors, and agency leaders responsible for ensuring AI is utilized in accordance with its stated obligations. Annual assessments of essential guidelines across the national security sector are mandated to adapt to advancements in AI. Whether these reviews will be substantive or merely symbolic remains uncertain.