U.S. Army AI Token Exhaustion and Push for Autonomous Weaponry

In June, the U.S. Army’s Combat Capabilities Development Command received an internal notice that its Ask Sage AI platform had exhausted the Department of Defense’s annual token allocation. Employees had been granted at least 200,000 tokens per month, with automatic top‑ups, but the pool of 100 million tokens—used for text and image generation across models such as Gemini, Llama and ChatGPT—was depleted within a year, prompting a reinstatement of usage limits and uncertainty about renewal after Oct. 1. Ask Sage, accredited for handling Controlled Unclassified Information, has been employed for tasks ranging from reclassifying personnel descriptions to broader enterprise workflows.

Concurrently, the Washington Post highlighted a broader U.S. push toward AI‑driven armed robots. A company called Foundation released footage of its “Phantom” robot carrying a mortar shell, exemplifying a shift toward autonomous weapons that can make rapid battlefield decisions without prior human approval. The Department of Defense is revising policies to accommodate lethal AI systems, while the Senate debates a security bill encouraging such technologies. Critics warn that unchecked deployment could destabilize future combat norms.