The Pentagon just approved 7 AI companies for classified DoD networks. One was excluded by choice. Pick your sensitivity level to see which providers you can actually use.
Select your sensitivity level
Why Anthropic said no
Anthropic was already widely used across the DoD. They were not excluded for poor performance. They refused to remove guardrails that limit weapons targeting and mass surveillance. The Pentagon called that a liability. Anthropic called it a line they would not cross.
First major AI company to publicly sacrifice a government contract for principles.
Building for government clients? AI provider selection is now a compliance decision, not just a capability one. Document it like you document your data privacy policy.
Anthropic is the only top-tier model that drew a public line on military use. Whether you agree or not, that is a different kind of moat.
For personal and business use, all major providers are cleared. The restrictions only kick in at government / classified sensitivity levels.
Data sources: Pentagon AI contract announcements (May 2026), published AI provider compliance pages. This tracker reflects publicly available policy information and is not a legal compliance opinion.