The US Government Just Put Claude on the AI Terror Watchlist

The US government reportedly looked at Anthropic’s Claude and went “national security supply-chain risk,” which is bureaucrat for “get this thing out of my house before it steals my silverware.”

That tag isn’t a mean Yelp review. It’s the kind of label that makes federal agencies and contractors start migrating like it’s a hurricane evacuation order, except the hurricane is a chatbot and the traffic is procurement.

Anthropic is fighting it in court, because nothing says “trust our model with your sensitive work” like a legal battle over whether the US government can treat you like a compromised vendor.

And “supply-chain risk” is the spookiest phrase in the whole tech economy. It doesn’t mean Claude wrote a rude poem. It means some agency thinks your dependencies, partnerships, hosting, training data, or access paths could become a backdoor on a bad day.

Meanwhile, every defense-adjacent contractor is now learning the most American lesson possible: you can build the coolest product on Earth, but if procurement thinks you’re “a risk,” you’re basically a banned book with a better UI.

This is the part where the winners aren’t “the best models.” It’s whoever gets cleared, certified, and stapled into federal buying workflows forever—because once the government picks a vendor, it clings like a tax lien.