The Anthropic-Pentagon saga just got a whole lot more interesting — and a whole lot more absurd.
‘Very Close’
New court documents filed on March 20 reveal a remarkable detail: On March 4 — one day after the Pentagon formally designated Anthropic as a “supply chain risk” — Under Secretary Michael emailed Dario Amodei to say the two sides were “very close” on the disputed issues.
Let that sink in. The Pentagon labels Anthropic a national security risk. And the very next day, a senior official writes that they were practically in agreement.
The Two Red Lines
What were the sticking points? Anthropic had two conditions: Claude must not be used for mass surveillance of American citizens. And Claude must not be deployed in fully autonomous weapons systems. Two demands that — let’s be honest — sound pretty reasonable.
The Pentagon disagreed. Defense Secretary Pete Hegseth designated Anthropic a supply chain risk. All federal agencies and contractors now have 180 days to remove Anthropic’s products from their systems.
Pushing Back
In two sworn declarations, Sarah Heck (Anthropic’s Head of Policy) and Thiyagu Ramasamy (Head of Public Sector) challenged the government’s claims. Their core argument: the case rests on technical misunderstandings and issues that were never actually raised during the months of negotiations.
Support has come from unexpected corners: Microsoft filed an amicus brief backing Anthropic. So did retired military leaders and — as we covered last week — Catholic ethicists.
Tomorrow It Gets Real
On March 24 — that’s tomorrow — Judge Rita Lin will hear the case in San Francisco. Anthropic is asking for a preliminary injunction. The decision could have far-reaching implications, not just for Anthropic but for the entire question of how AI gets used in defense.
My take: these new documents seriously weaken the Pentagon’s position. If you were nearly aligned and then killed the deal for political reasons — that’s a tough sell, even in court.
Sources: