The Pentagon announced in May 2026 that it has struck classified agreements with seven AI companies — OpenAI, Google, Microsoft, Amazon, Nvidia, Elon Musk’s xAI, and startup Reflection — allowing their tools to be used in classified military settings. Notably absent from the list is Anthropic, a company the Defense Department had previously worked with, after declaring it a supply-chain risk.
The agreements build on existing arrangements. OpenAI and xAI had already reached deals with the Pentagon for the “lawful” use of their AI systems. According to a report from The Information, Google struck a similar agreement, while The Wall Street Journal noted that Microsoft and Amazon already have “deep relationships with the Pentagon.” Contracts with Nvidia and Reflection are described as new.
In its announcement, the Defense Department said the agreements will allow for the “lawful operational use” of the companies’ systems, with the stated goal of “establishing the United States military as an AI-first fighting force.”
Emil Michael, the Defense Department’s chief technology officer, addressed Anthropic’s exclusion in a Friday interview with CNBC, confirming the company remains a supply-chain risk. He described Anthropic’s security-focused AI model, Mythos, as a “separate national security moment,” adding: “We have to make sure that our networks are hardened up, because that model has capabilities that are particular to finding cyber vulnerabilities and patching them.”
The deals signal a significant expansion of the U.S. military’s use of commercial AI tools in sensitive environments. The exclusion of Anthropic — despite its previous role in classified work — suggests the Pentagon is applying new scrutiny to its AI supply chain, which could affect how AI companies pursue future government contracts.
Source: The Verge