Pentagon Tests AI Models to Replace Claude

Pentagon began testing AI models with 25 power users in early March to find alternatives to Anthropic’s Claude after supply‑chain risk designation.
Defense Secretary Pete Hegseth labeled Anthropic a supply‑chain risk, prompting a six‑month phase‑out and a court challenge claiming billions in lost revenue.
Pentagon signed agreements with vendors like OpenAI and Google to deploy AI tools on classified networks within a month‑long testing window.
Undersecretary Emil Michael said talks with Anthropic are suspended and new competitor models are expected monthly, with performance varying by prompts.