Pentagon Signs AI Contracts
The Pentagon signs contracts worth up to $200 million each with Anthropic, OpenAI, Google, and xAI to deploy AI models across military operations, beginning months of negotiations over terms of use.
In the summer of 2025, the Pentagon signed contracts valued at up to $200 million each with four major AI companies. The contracts aimed to bring frontier AI models into military operations across weapons development, intelligence collection, and battlefield support.
The “All Lawful Purposes” Demand
The Pentagon’s core requirement was that all models be available for “all lawful purposes.” Under Secretary Emil Michael later acknowledged these contracts “lacked initial specificity” on terms of use — an ambiguity that became the fault line of the subsequent dispute.
Divergent Responses
- OpenAI, Google, and xAI agreed to lift consumer-facing guardrails for military use
- Anthropic refused to agree fully, insisting on hard limits against mass surveillance and autonomous weapons
Claude on Classified Networks
Claude was reportedly the first AI model brought into the Pentagon’s classified networks, deployed through Palantir’s Artificial Intelligence Platform (AIP). This early access made the subsequent dispute particularly consequential — Claude was already integrated into sensitive military infrastructure.