The next round in the legal battle between Anthropic and the Trump administration has begun. On March 17, the US Department of Justice filed its defense brief — and the message is clear: Anthropic’s blacklisting was justified and lawful.
What happened?
Quick recap: Defense Secretary Pete Hegseth designated Anthropic a ‘national security supply chain risk’ on March 3 — after the company refused to remove safety guardrails against Claude being used for autonomous weapons and domestic surveillance. On March 9, Anthropic sued, calling the designation ‘unprecedented and unlawful.’
Now the other side has responded.
The DOJ’s arguments
The government argues on multiple levels: First, this isn’t a free speech case — it’s a contract negotiation in the realm of national security. Second, the Pentagon has the right to exclude suppliers that don’t meet its security requirements. And third, Anthropic is ‘unlikely to succeed’ with its lawsuit — a legal signal that the government is pushing for dismissal.
What’s at stake
This is about more than a single contract. The ‘supply chain risk’ designation could be extended to the entire federal government. Anthropic itself warns it could reduce 2026 revenue by ‘multiple billions of dollars.’ The company has filed a second lawsuit in a Washington, D.C. appeals court.
At the same time, Anthropic is growing despite — or because of — the crisis. Annual revenue is above $20 billion, and the Pentagon controversy has given Claude a sympathy boost among users who are critical of OpenAI’s willingness to work with the military.
My take
This lawsuit will set a precedent. Can the government punish a tech company for setting ethical boundaries? Or does the Pentagon have the right to choose partners that cooperate unconditionally?
For Anthropic, it’s an existential question — but also one that paradoxically earns the company more public support than any marketing campaign ever could. The answer won’t just determine Anthropic’s future — it’ll set the framework for how much pressure governments can put on AI companies to comply.
Sources: