US Military Will Integrate Elon Musk's Grok AI into Classified Systems

The Pentagon has agreed with Musk’s company xAI to allow Grok to be used in highly sensitive defense systems, including those handling intelligence analysis, weapons development, and battlefield planning. This marks a shift from previous practice, in which Anthropic’s Claude was the only AI model widely deployed on classified networks.

Grok has faced global backlash for generating inappropriate and sexually explicit content, including deepfake-style images of people without consent, and governments and lawmakers have criticized its safety and reliability.

The move comes amid a dispute between the Pentagon and Anthropic over terms of use for its Claude model. The Defense Department has asked Anthropic to make Claude available for “all lawful purposes,” including potential military applications like mass surveillance and autonomous weapons systems. Anthropic has resisted those conditions, arguing that its safety restrictions are necessary, even as it remains involved in some defense operations.

By contrast, xAI agreed to the Pentagon’s requirement that Grok be usable for any “lawful” military function, clearing the way for its inclusion in defense AI infrastructure. However, U.S. officials acknowledge that Grok is generally regarded as less advanced than Claude and that transitioning key duties away from Claude could be technically challenging.

The Pentagon is also in talks with other major AI developers, including OpenAI and Google’s Gemini, as it seeks alternatives amid ongoing tensions with Anthropic.