Pentagon officials say a sudden realization of operational dependence on Anthropic’s models triggered an urgent reassessment of defense AI supply chains, a top official recounted on a podcast. That scramble followed a public split with Anthropic and a presidential order limiting federal use of the company’s models. > The dispute rippled into industry and research communities: Caitlin Kalinowski, OpenAI’s robotics lead, resigned citing concerns about surveillance and lethal autonomy after the company struck its own Defense Department deal. Universities and research labs that partner with defense contractors now face tightened export, procurement and ethics scrutiny, and administrators must review compliance clauses in sponsored‑research agreements.