A return through civilian agencies
The Pentagon declared Anthropic a supply-chain risk in March after the company refused to remove restrictions on mass surveillance and autonomous weapons. Donald Trump then ordered all agencies to stop using Claude. The military kept using it during the war in Iran, but later took it out of service.
The legal fight has not settled the dispute. In late August, a judge in San Francisco ruled that one of the two classifications was an unlawful act of retaliation. In late September, an appeals court in Washington upheld the other.
Anthropic’s government offering includes:
The unresolved question
Anthropic CEO Dario Amodei’s dinner with Trump on Sunday may signal that tensions are easing, but the change could be temporary. I think the more important test is whether agencies can keep using Claude through this political and legal uncertainty. The offer provides administrative controls and local storage; it does not settle the disagreement over which uses Anthropic will permit.
That leaves a distinction between access and trust. Agencies can buy the same service as corporate customers, but the Pentagon dispute shows that government access can still turn on questions a procurement contract cannot resolve.
Daily AI news
Every day we pick what actually matters in AI and explain it plainly — no hype, no filler. Subscribe if you want to follow where the industry is going.
Only what matters — every day
Follow on X