Court Upholds Pentagon Ban on Anthropic Over Military AI Usage Limits
A federal court backed the Pentagon ban on Anthropic, locking Claude out of defense contracts over autonomous weapons limits.
Policy · Source: Hacker News
What happened
A federal appeals court in Washington upheld the Pentagon decision to blacklist Anthropic. The 2-1 ruling keeps the AI company labeled as a supply chain risk. This designation completely blocks the U.S. military from using Claude models. It also prevents defense contractors from using Anthropic technology in their agency work. The court delayed the immediate effect of the ruling to give Anthropic time to seek a rehearing or petition the Supreme Court.
The conflict stems from a massive deal signed last year. Anthropic secured a $200 million contract with the Pentagon in July 2025. Talks collapsed in September when negotiating Claude deployment on the military GenAI platform. The Department of Defense demanded unfettered access for all lawful purposes. Anthropic demanded guarantees that Claude would not be used for fully autonomous weapons or domestic mass surveillance. Defense Secretary Pete Hegseth accused Anthropic of trying to seize veto power over military operations. Hegseth also claimed overly constrained models could shut down unexpectedly or be manipulated.
The legal battle has turned into a bitter political feud. President Trump recently mocked Anthropic CEO Dario Amodei on Truth Social, claiming his administration stopped AI people from doing bad things. Amodei was also excluded from a recent White House state dinner for the Chinese President after calling for an industry slowdown. Anthropic previously won a related case in a San Francisco federal court last month, but this new D.C. ruling upholds a second designation that keeps the ban alive.
Key facts
- 2-1 — Appeals court decision ratio upholding the Pentagon ban
- $200 million — Value of the July 2025 Pentagon contract Anthropic signed before talks collapsed
- March 2026 — When the Department of Defense originally labeled Anthropic a supply chain risk
Why it matters
If you build AI for the federal government, your terms of service do not matter. The Pentagon demands absolute, unfettered access to the underlying models. Trying to enforce ethical guardrails on defense contracts will get you blacklisted and labeled a national security risk. Defense contractors using Claude in their own software stacks now face a massive headache. They have to rip Anthropic out of their pipelines or risk losing their government deals.
This ruling effectively splits the AI industry into two distinct camps. You either build unrestricted models for the military, or you get locked out of massive federal budgets. Competitors who do not care about autonomous weapons restrictions will gladly take Anthropic market share. The decision also sets a dangerous legal precedent. The government can now use the Supply Chain Security Act to punish American companies purely for their internal safety policies.
For builders
Defense contractors must rip out Claude
The ban directly applies to defense contractors working with the agency. If you sell software to the Pentagon, you cannot use Anthropic models in your pipeline. You must migrate to OpenAI or open-source alternatives immediately or lose your contracts.
Ethical guardrails kill federal contracts
Anthropic lost a $200 million deal because they refused to allow fully autonomous weapons. Founders must choose between strict AI safety principles and lucrative defense revenue. The military will not accept vendor-imposed operational limits on their technology.
Supply chain laws weaponized against AI
The government used the Supply Chain Security Act to blacklist an American company over policy disagreements. This creates massive regulatory risk for AI startups. Your corporate safety guidelines could legally classify you as a national security threat.
My take
You cannot sell technology to the military and tell them how to fight. Anthropic tried to have it both ways by taking a massive defense contract while dictating military strategy. If you want government money, you have to play by their rules. Your safety charter means nothing to the Pentagon.
Original reporting: Hacker News. This is my rewrite and opinion.