Pentagon drops Anthropic for competitors

The Pentagon has shifted two-thirds of its AI tasks from Anthropic's models to those of other companies. The move follows a dispute over Claude's acceptable use policy. In early 2026, the US Department of Defense required Anthropic to remove limits on mass surveillance and autonomous weapons use, which Anthropic CEO Dario Amodei refused.

In March 2026, the Pentagon attempted to blacklist Anthropic, but the decision was blocked by a court. Some $200 million in contracts were canceled. As a result, the Pentagon expanded its partnerships with OpenAI, Google, and Microsoft.

OpenAI emerged as the primary beneficiary, having already removed its direct ban on military use of its technologies in 2024.