Vance warns US military to keep AI from life-and-death calls

Vance’s Air Force Academy speech is one of the clearest high‑level statements yet from the Trump administration drawing a red line around AI in warfare. By explicitly tying his position to Pope Leo XIV’s encyclical and insisting that “decisions over life and death must be made by humans and not machines,” he’s signalling to the Pentagon, contractors and allies that full autonomy in lethal systems is politically radioactive, even as battlefield AI proliferates. ([washingtonpost.com](https://www.washingtonpost.com/politics/2026/05/28/vance-says-military-should-never-let-ai-make-life-and-death-decisions/?utm_source=openai))

In the practical race to AGI, this doesn’t slow model development directly—but it could reshape demand signals. If US doctrine hardens around “human in the loop” requirements, labs and integrators may focus more on decision‑support, targeting assistance and logistics optimisation rather than fully autonomous kill chains. That still requires cutting‑edge models, but with more emphasis on auditability, control and human‑factors tooling. It also sets up a stark contrast with adversaries who may be less constrained, potentially increasing pressure inside the US security establishment to relax norms later.

For now, Vance is helping mainstream the idea that frontier AI’s most consequential uses must be norm‑bound and morally legible. Whether that becomes lasting policy or just rhetoric in the face of a fast‑moving arms race will matter a lot for how AGI gets militarised.