Pentagon Reaches Confidential AI Cooperation Agreements with OpenAI, Google, and Nvidia, Excluding Anthropic
The Pentagon has reached agreements with OpenAI, Google, Microsoft, Amazon, Nvidia, Elon Musk's xAI, and startup Reflection, allowing them to use their AI tools on classified networks and in classified scenarios. The Department of Defense has excluded Anthropic from the new round of cooperation, citing it as a "supply chain risk."
These new agreements are an expansion of previous Pentagon collaborations with OpenAI and xAI, which were limited to "legitimate purposes" for military AI use. Reports indicate that Google has also reached a similar agreement with the Pentagon, allowing its AI systems to be used in specific defense scenarios. The Wall Street Journal notes that Microsoft and Amazon already have "deep relationships" with the Pentagon, while Nvidia and Reflection are new additions to the cooperation list.
In contrast, Anthropic previously signed a $200 million contract with the Pentagon to process classified materials, but disagreements arose during the collaboration regarding "red lines" for AI in large-scale domestic surveillance and fully autonomous weapons. Anthropic refused to relax related restrictions for the Department of Defense, ultimately leading to controversy and its products being placed on the federal government's prohibited procurement list. Anthropic subsequently sued the U.S. federal government and obtained a temporary injunction in legal proceedings.
Department of Defense Chief Technology Officer Emil Michael told CNBC that Anthropic is still considered a supply chain risk, but also stated that its powerful security model, Mythos, constitutes "another level of national security event." He emphasized that, given the model's unique ability to identify vulnerabilities in network systems and complete patching, "we must ensure our own networks are fortified."
In a formal announcement, the Pentagon stated that the agreements with the seven AI companies will support the military's "legitimate operational use" of its systems and declared that this will "build the U.S. military into an AI-first fighting force."