Report from the Wall Street Journal
In Brief – The US Defense Department has finalized agreements with eight major US technology companies to deploy their artificial intelligence tools in classified environments. The companies are OpenAI, Google, SpaceX (via xAI), Microsoft, Amazon, Oracle, Nvidia, and startup Reflection AI. The deals reflect growing alignment between Silicon Valley firms and Pentagon AI requirements, particularly in the wake of Anthropic refusing a similar contract and being deemed a supply-chain risk. The agreements aim to give service members a wide range of AI models to work with. Existing corporate defense contractors like Microsoft, Amazon, and Oracle deepen their roles into AI services, while new partnerships with Nvidia and Reflection AI emphasize open-source AI models. The Pentagon’s push comes amid intensifying global AI competition, particularly with China’s development of open AI models, including for export.
Context – The second Trump Administration, with a cohort of tech industry backers, has pitched AI development as a national imperative to compete with China. In the wake of the Pentagon’s legal blowup with Anthropic, and in a throwback to the progressive Google employee activism of 2018 and 2019, more than 600 Google employees, including at its Deep Mind AI business, urged CEO Sundar Pichai to block the Pentagon from using Google AI in classified settings. The C-Suite response is clearly no thanks. The whole debate is another example of science fiction narratives filling an AI policy vacuum. Nobody knows how AI systems will evolve and everyone involved is too steeped in science fiction, even the AI models themselves. Killer AI robot concerns tanking defense contracts and job apocalypse fan fiction tanking SAAS stocks. Then there are conservative charges of Woke AI now being leveled at Anthropic, which is the ideological cousin of 2020 social media censorship. While the ability of the federal government to set AI standards for defense work is likely on solid legal ground, using the designation as a DoD supply-chain risk to blacklist an AI company from all federal contract work is less likely.