The Pentagon just handed classified military access to OpenAI, Google, Nvidia, Microsoft, Amazon, SpaceX, and startup Reflection AI. The goal: transform the US military into an 'AI-first fighting force.'

Notably absent: Anthropic, the Claude chatbot maker that refused to sign the Pentagon's 'any lawful use' clause. In response, the Defense Department labeled Anthropic a supply-chain risk last month, the first time an American AI company has received that designation.

The deals integrate these companies into the Pentagon's 'Impact Levels 6 and 7' classified networks. Translation: the same AI models powering your ChatGPT chats and Google searches will soon help the military 'streamline data synthesis' and 'augment warfighter decision-making in complex operational environments.'

The Pentagon is budgeting tens of billions for cutting-edge AI programs, including $54 billion requested just for autonomous weapons development. Individual deployment details remain classified, but the scope is massive: intelligence analysis, drone warfare, classified communications, and weapons systems.

Anthropic's exclusion stems from a months-long feud over AI safety guardrails. The company objected to Pentagon language allowing 'any lawful use' of its technology, citing concerns about mass surveillance and fully autonomous lethal weapons. When Anthropic refused to budge, the Pentagon struck back with the supply-chain risk designation.

One signatory stands out: Reflection AI, a two-year-old startup that hasn't released a public model yet. The company's mission is creating open-source AI to counter Chinese firms like DeepSeek. It's seeking a $25 billion valuation and has backing from Nvidia and 1789 Capital, where Donald Trump Jr. is a partner.

Defense Secretary Pete Hegseth unveiled this 'AI acceleration strategy' in January, promising to 'unleash experimentation, eliminate bureaucratic barriers' and ensure American military AI dominance. The Anthropic contracts suggest the Pentagon believes signing with safety-focused competitors will pressure the holdout back to the table.

The irony is stark: the one major AI company prioritizing safety guardrails gets banned as a national security risk, while everyone else signing 'any lawful use' clauses gets classified access. It's a clear signal from the Pentagon about what matters more in the AI arms race.