The Pentagon has struck deals with major tech giants like OpenAI, Google and Nvidia to use their AI tools in classified settings. However, Anthropic, known for its powerful security model Mythos, was left out due to concerns over supply-chain risks.
These agreements follow existing partnerships with companies like Microsoft and Amazon who already have deep relationships with the Pentagon. Nvidia's and Reflection's contracts are new.
Despite Anthropic’s $200 million deal for handling classified materials, it refused to relax its strict limits on domestic surveillance and autonomous weapons, leading to a dispute that saw its products barred from federal use. A temporary injunction followed after Anthropic sued the government.
Emil Michael, the Pentagon's chief tech officer, highlighted Anthropic’s security model as a significant risk, particularly due to its advanced cyber vulnerability detection capabilities.
The Pentagon asserts these deals will ‘establish the United States military as an AI-first fighting force,’ but critics wonder if exclusion could set back ethical defence technology advancements.
