The Intercept Reveals Pentagon Contracts Worth Up to $200M Each with OpenAI, Anthropic, Google, and xAI

My take: This was expected, but seeing the numbers makes it more concrete: The Intercept published more than 400 pages of U.S. Department of Defense contracts, obtained through FOIA litigation, showing that OpenAI, Anthropic, Google, and xAI each signed agreements in July 2025 with a ceiling of $200 million per company to develop artificial intelligence prototypes inside classified military systems.

The contracts center on "agentic" workflows, meaning AI systems that operate with some autonomy, integrated into platforms like Advana and Maven Smart System to support everything from combat planning to payroll administration. What stands out is that the four most relevant labs in the sector, whose models millions of people use every day, are now also part of the U.S. national defense ecosystem.

For the professional or business relying on these tools, the news does not change their practical value, but it does reinforce something worth keeping in mind: understanding who has access to your data and what the companies behind your tools do with their models is not just a curiosity, it is part of using technology with intention.

How much do you really know about the companies behind the AI tools you use at work?