Overview
- President Donald Trump ordered federal agencies to stop using Anthropic immediately with a six‑month phase‑out for the Defense Department and other users, according to agency notices and public statements.
- Multiple outlets report that Anthropic’s Claude supported U.S. targeting, simulations and intelligence in an operation against Iran that coincided with the reported death of Ali Khamenei, while CENTCOM declined to detail specific systems.
- OpenAI announced an agreement to deploy its models inside classified Pentagon networks under stated guardrails, and the State Department is migrating its internal chatbot to GPT‑4.1 as agencies unwind Anthropic tools.
- Anthropic refused Pentagon demands to drop restrictions on mass domestic surveillance and fully autonomous weapons, and the company says it will challenge any supply‑chain risk designation in court.
- Anthropic reported global service disruptions on March 2 with frequent errors affecting Claude.ai log‑in and session flows, as outage trackers showed rising incident reports.