The U.S. military has finalized a deal with seven major AI labs, including Google, Microsoft, AWS, NVIDIA, OpenAI, Reflection, and SpaceX, to integrate their artificial intelligence into classified systems, while excluding Anthropic due to its refusal to lift restrictions on AI’s use in warfare. This development marks a significant step in the Pentagon’s rapid integration of generative AI into its operations, as demonstrated by the GenAI.mil platform, which has been utilized by over 1.3 million personnel in just five months. Additionally, the collaboration with multiple firms aims to prevent reliance on a single AI provider, ensuring greater operational flexibility.
AWS: Amazon Web Services (AWS) leads in cloud computing, offering scalable infrastructure optimized for AI and machine learning workloads. AWS provides compliant services for government and military use cases. This deal allows AWS to deploy AI capabilities directly on classified DoD networks.
Google: Google develops frontier AI models like Gemini and provides government-compliant cloud services through Google Cloud. It recently integrated its latest Gemini models into the Pentagon’s GenAI.mil platform. Google is now authorized to run its AI technologies on classified military systems.
NVIDIA: NVIDIA produces GPUs critical for AI acceleration and develops open-source models like Nemotron. Its hardware underpins global AI development and deployment. NVIDIA joined the agreements to make its AI models available on Pentagon classified systems.
OpenAI: OpenAI builds advanced generative AI systems like ChatGPT, with custom versions for secure government environments. It previously deployed ChatGPT on GenAI.mil for defense teams. OpenAI agreed to unrestricted lawful use of its models on military classified networks.
SpaceX: SpaceX engineers reusable rockets, satellite constellations like Starlink, and autonomous systems for space and defense applications. The company expands into AI-enabled operations for military relevance. SpaceX partnered with the Pentagon to provide AI access on secure classified platforms.
Pentagon: The Pentagon is the headquarters of the US Department of Defense, overseeing military strategy, operations, and technological integration for national security. It operates GenAI.mil, its official generative AI platform, to foster an AI-first workforce across unclassified and classified environments. The Pentagon expanded classified AI agreements with seven firms to enable deployment of their models for lawful operational uses on secure networks.
Anthropic: Anthropic creates AI models prioritizing safety, alignment, and built-in ethical constraints. The company faced tensions with the DoD over mandatory safety guardrails for military AI applications. Anthropic was explicitly excluded from the classified agreements as a designated supply-chain risk.
Microsoft: Microsoft delivers Azure cloud platforms and AI tools tailored for defense and enterprise applications. The company supports DoD initiatives with secure AI infrastructure. Under the new agreement, Microsoft enables its AI models to operate on Pentagon classified networks.
Reflection: Reflection AI is a startup specializing in open-source large language models, such as Reflection 70B. The company focuses on accessible, high-performance AI technologies. Reflection secured approval to integrate its models into the Pentagon’s classified networks.
Safety Dispute: Anthropic’s exclusion stems from its refusal to waive restrictions on AI use in warfare scenarios.
AI Acceleration: GenAI.mil showcases the Pentagon’s rapid push to embed generative AI into daily military workflows.
Vendor Diversity: Multiple firm agreements safeguard against AI provider dependency and ensure operational flexibility.