Pentagon signs classified AI deals with Nvidia, Microsoft, and AWS after ejecting Anthropic over safety limits

The Pentagon has expanded its classified artificial intelligence partnerships by signing deals with seven companies, including major tech firms Nvidia, Microsoft, and Amazon Web Services (AWS). These agreements grant the Department of Defense access to the companies’ AI technologies and tools for “lawful operational use” within classified military networks. This move comes amid efforts to accelerate the military’s adoption of AI with enhanced capabilities and represents a strategic step to develop an AI-first approach across defense operations. Alongside these companies, others such as OpenAI, Google, SpaceX, Reflection AI, and xAI have also been included in this classified AI collaboration.

Significantly, Anthropic, a prominent AI company known for its Claude AI model, was excluded from these agreements after refusing to lift certain safety restrictions demanded by the Pentagon. The Trump administration had previously banned Anthropic’s tools from military and government use due to concerns over safety guardrails. The Pentagon’s decision to replace Anthropic reflects a prioritization of operational flexibility and fewer limits on AI usage in sensitive defense applications.

This development underscores the increasing militarization of artificial intelligence and the importance of partnerships between the defense sector and leading AI developers. It highlights how national security considerations are shaping AI deployment policies, balancing between innovation, safety, and operational efficacy. By enlisting industry leaders like Nvidia, Microsoft, and AWS, the Pentagon aims to leverage cutting-edge AI while ensuring compliance with classified operational standards.