The Pentagon has reached agreements with OpenAI, Google, Microsoft, Amazon, Nvidia, Elon Musk's xAI, and the startup Reflection, allowing the use of these companies' artificial intelligence tools in classified networks and scenarios. At the same time, the U.S. Department of Defense excluded Anthropic, which had previously been used to process classified information, from a new round of cooperation on the grounds that it was identified as a "supply chain risk."
These new agreements further expand on the Pentagon’s previously signed AI military cooperation with OpenAI and xAI, which is only used for “legitimate purposes.” Reports indicate that Google has reached a similar agreement with the Pentagon to allow its AI systems to be used in specific defense scenarios. The Wall Street Journal pointed out that Microsoft and Amazon already have "deep relationships" with the Pentagon, while Nvidia and Reflection are new faces on the list of cooperation.
In sharp contrast, Anthropic signed a $200 million contract with the Pentagon to process classified materials. However, during the cooperation process, the two parties had serious differences around the "red line" of AI in large-scale domestic surveillance and fully autonomous weapons. Anthropic refused to relax relevant restrictions for the Department of Defense, ultimately causing controversy and its products being banned from federal government procurement. Anthropic subsequently sued the U.S. federal government and obtained a temporary injunction during the judicial process.
Defense Department Chief Technology Officer Emil Michael told CNBC that Anthropic is still considered a supply chain risk, but said its powerful security model, Mythos, poses "another level of national security incident." He emphasized that given the model's special ability to discover network system vulnerabilities and complete patching, "we must ensure that our own networks are hardened."
In a formal announcement, the Pentagon said the agreement with the seven AI companies would support the U.S. military’s “legitimate combat use” of its systems, declaring that it would “build the U.S. military into an AI-first combat force.”