Appeals court upholds Pentagon's supply chain risk blacklist of Anthropic, blocking Claude from DOD and its contractors
- A D.C. Circuit panel ruled 2-1 on Friday to uphold the Department of Defense's March supply chain risk designation of Anthropic, blocking the U.S. military and any defense contractor working with the agency from using Claude models.
- Judge Gregory Katsas wrote the majority opinion, joined by Judge Neomi Rao, holding that the Department had ample support for finding that continued integration of Claude into its information systems presented a statutorily covered national-security risk. Judge Karen LeCraft Henderson dissented.
- The ruling rejected Anthropic's argument that the DOD's ban was arbitrary, unauthorized, and unconstitutional, after the labeling followed negotiations over how the military could use Claude models that spiraled out of control.
- Anthropic sued the Trump administration in March in both a San Francisco district court and the D.C. Circuit, because the DOD relied on two distinct designations to justify the action and they had to be litigated in two separate courts.
- An Anthropic spokesperson said the company is considering all options including further review, while Trump has repeatedly attacked CEO Dario Amodei on Truth Social and Amodei was not invited to the White House state dinner for Xi Jinping on Thursday.
Hacker News opinions
I wonder if this is going to start being abused soon. If I were Palantir or any other GOP aligned company, I'd be completely against this. What stops a Democratic president from doing the same thing and destroying them?
I'm worried why so many big companies aren't worried about the shoe being on the other foot.
How would this destroy Anthropic, exactly? The Supreme Court, I guess. What makes you think they'd stop it while Trump appointees hold the majority there?
Democrats don't control the courts. That's the actual check here.
Palantir isn't concerned. It's LLM agnostic and holds IL5, IL6, and ICD 503 compliance. Their whole platform is built around eliminating this kind of risk.
If the CEO of any other defense contractor did this, the board would fire them on the spot.
Not ones whose defense customer base is dwarfed by commercial. No defense contractor CEO has built a company this large this fast, and nobody is thinking about firing Dario.
I still don't fully understand what this was about. It reads like DoD wanted unrestricted access to Anthropic's models, Anthropic refused, so the Pentagon said it wouldn't use them at all. Isn't that what Anthropic wanted?
No, because it also means no companies working with DoD can use Anthropic models for any purpose. The Pentagon is basically bullying contractors into dropping them too.
Anthropic wants to sell to the military, but on its own terms and with limits on what it can be used for. Apparently that disagreement got escalated into a supply chain risk label.
It came down to one point: Anthropic wanted to review actions affecting human life. Deferring that judgment to a private company is a regression from centuries of refined military practice.
Anthropic just didn't want AI making battlefield kill decisions without a human in the loop because the models are too unreliable. That is not a supply chain risk, and the objection was confirmed when the military hit a school in Iran after an AI error.
And crucially, DoD already agreed to Anthropic's terms through the normal procurement process and made the purchase. Much later the Secretary of Defense threw a political fit and tried to rewrite the terms.
You're skipping the supply chain risk designation, which is an extraordinary and punitive measure. Without it this would be a nothingburger, since 'government picks vendor Y instead of X' isn't news.
Supply chain risk is a massive over-reaction here, but I get why the government doesn't want to commit to Anthropic-controlled software for critical military work if it can't fully rely on Anthropic support.
That's mobster-loyalty logic. Also nobody disputes the Pentagon can write whatever RFQ it wants. The real subject is using national security regulatory power to coerce Anthropic into selling a product it doesn't want to sell.
This smells like corruption to me. OpenAI hacked several prominent entities and keeps doing business as usual, while Anthropic putting guardrails on military AI use is treated as the national security threat. I'm moving my money back to Anthropic.