Overview
- Defense Secretary Pete Hegseth told CEO Dario Amodei to accept “all lawful” military uses of Claude by 5 p.m. Friday or face compelled cooperation, according to multiple reports and a senior Pentagon official.
- Officials warned they could invoke the Defense Production Act, label Anthropic a supply‑chain risk, and terminate contracts if the company refuses to drop guardrails.
- Anthropic says it will not enable mass domestic surveillance or fully autonomous lethal targeting, though it agreed in December to missile and cyber defense uses and says talks remain in good faith.
- xAI reached a deal to allow its Grok system on classified networks for any lawful purpose, increasing pressure on Anthropic, which has been uniquely integrated via Palantir.
- On Tuesday Anthropic revised its Responsible Scaling Policy to remove a prior pledge to pause development without a safety lead, while Sens. Elizabeth Warren and Andy Kim condemned the Pentagon’s threat as an abuse of the DPA.