153: US Military Used Claude to Target Iran Despite Trump's Ban

Get AI-Ready with Kyle’s 5-Day Challenge: https://aiwithkyle.com/join

Learn how to become an AI trainer - Live Webinar: https://aiwithkyle.com/ai-workshop-series/webinar

Kyle’s AI Canon — Curated list of AI learning essentials: https://aiwithkyle.com/ai-canon

Subscribe and turn on notifications to catch the next live stream: https://www.youtube.com/channel/UChlLglbHDASnoGkbjDeHnQg

This video is from the ‘AI with Kyle Live Stream’

First aired: 2 March 2026

Summary: The same AI model you're using to write emails and debug code was reportedly used by the US military to select bombing targets in Iran this weekend. Not a drill, not a hypothetical. And the detail that should genuinely make you pause is this: the government doesn't have a secret, more powerful version of Claude sitting in a classified bunker somewhere. Ethan Mollick made the point well. They're using the same model you get for $20 a month, probably an older version, because new models have to go through procurement before deployment. That's the actual state of AI in modern warfare right now.

What that means in practice is worth sitting with. When Claude is a little too agreeable, a little too confident, prone to the occasional hallucination, that's fine when you're polishing a draft or writing a proposal. When it's informing decisions about military targets, the stakes are rather different. The AI arms race isn't some distant future concern. It's already here, already running, and it's built on the exact same infrastructure the rest of us use every day.

—— Time Stamps ——

0:00 Introduction: Trump Enters the Anthropic War 0:41 Analysis of Donald Trump's Truth Social Post 2:11 Anthropic's Red Lines: Surveillance and Weapons 3:34 Pete Hegseth Designates Anthropic a Supply Chain Risk 5:58 Impact on AWS, Palantir, and Tech Investors 7:53 Why Anthropic Can't Move Operations to Europe 8:56 US Military Uses Claude in Iran Strikes 11:09 Ethan Mollick on Government AI Access 12:41 Military Training and AI Hallucinations 17:05 Ethical Divide: Domestic vs Foreign AI Use 18:22 OpenAI Reaches Deal with the Pentagon 19:52 Comparing OpenAI and Anthropic Safety Guards 21:14 Drone Warfare and Autonomous Decision Making 22:41 Centralized vs Decentralized AI Control 24:05 Shifting Regulations on Human Control in AI 25:35 Posse Comitatus Act and Domestic Enforcement 29:44 OpenAI's FAQ on Military Safety Agreements 31:10 OpenAI vs Anthropic: Well-Oiled Machine vs Scrappy 33:49 Military Procurement and AI Model Versions 39:28 Claude's Growing Market Share vs ChatGPT 42:12 Scaling Challenges and AWS Risk 46:10 China's AI Progress: Manus and Agentic Tech 49:50 How to Stay Updated with the Newsletter

—— More Useful Resources ——

Find everything else at https://aiwithkyle.com/