U.S. Military Tests AI Kill Chain in 30-Nation Morocco Drill

3 articles · Updated · CBS New York · May 29

African Lion 2026 in Morocco became a live test bed for U.S. AI warfare tools, with troops using autonomous vehicles, armed drones and AI-assisted command systems during the 30-nation exercise.
Project Maven, built by Palantir, cut one targeting decision from two or three hours to three minutes, according to a U.S. Army officer, while a human still approved the artillery strike.
Multiple sources said Maven’s operator interface relied on Anthropic’s Claude model, showing the company remains embedded in Pentagon workflows despite Defense Secretary Pete Hegseth calling it a national-security supply-chain risk.
Two robots replaced about 40 soldiers in one breaching drill, an officer said, underscoring the Army’s push to use autonomy to keep troops off the front line and speed battlefield action.
Senior commanders and soldiers said fully autonomous lethal decisions remain ethically troubling, but warned the U.S. risks falling behind adversaries if it does not keep adopting the technology.

With AI making kill decisions in minutes, is 'human-in-the-loop' becoming a dangerous illusion for future warfare?

Will Silicon Valley's ethics bend to the Pentagon's need for an AI edge against global adversaries?