Pentagon issued an ultimatum to Anthropic: provide full AI access by Friday or face sanctions.

Brief overview of the situation

The U.S. Department of Defense (DoD) forced Anthropic to open full access to its AI model within a week. If it does not, the company will face serious consequences.

What happened
1. DoD inspection

At a meeting with Secretary of Defense Pete Hegseth, Dario Amodei – head of Anthropic – received a warning: either the company would be declared a “supply‑chain threat” (a status usually applied to foreign adversaries), or the DoD would invoke the Defense Production Act (DPA).

2. What the DPA means

- It allows the president to compel companies to prioritize and expand production in the interests of the military.
- In the past, this law was used to force General Motors and 3M to produce ventilators and masks during the pandemic.

Why Anthropic is resisting
- Protecting civil rights – the company does not want its technology used for mass surveillance of Americans or employed in fully autonomous weapons.
- Constitutional limits – they require that military applications be governed by federal law, not private contractor policy.
- No easing of rules – Anthropic refuses to relax restrictions on how its models can be used.

Current status
- Anthropic remains the only leading AI lab with access to DoD classified materials, and the department has no backup option yet.
- Unverified reports say the Pentagon has already signed agreements with xAI to use the Grok model in closed systems.

Experts believe that the lack of alternatives may explain the “aggressive” stance of the Department of Defense.