Anthropic Rejects Pentagon AI Deal

The Pentagon's demand for "all lawful purposes" access to Anthropic's AI was the deal's breaking point. This broad language would have granted the Department of Defense wide latitude, which conflicted with Anthropic's foundational principles against mass surveillance of American citizens and the deployment of fully autonomous weapons. The company offered a compromise to allow the processing of classified material under FISA but held firm on prohibiting the warrantless collection of commercial data on ordinary people. This ethical stance is rooted in the company's 2021 founding by former OpenAI executives, including siblings Dario and Daniela Amodei, who left the prominent AI lab over safety concerns. Anthropic was established as a Public Benefit Corporation, legally structured to prioritize positive social impact over profits and to focus on creating steerable and interpretable AI systems. This mission is codified in their "Constitutional AI" framework, which bakes in a set of guiding principles to align the AI's behavior with human values. In response to the impasse, the U.S. government designated Anthropic a "supply chain risk," a label typically reserved for foreign adversaries and never before applied to a U.S. company. This move effectively orders all military contractors and partners to cease commercial activity with the firm, prompting Anthropic to challenge the designation in court. Within hours of the Anthropic deal collapsing, OpenAI secured its own agreement with the Pentagon. OpenAI stated its deal includes prohibitions on mass surveillance and autonomous weapons, functionally similar to what Anthropic had requested. The key difference appears to be in enforcement: Anthropic insisted on explicit contractual restrictions, while OpenAI relied on a combination of contract language, a cloud-only deployment architecture, and ongoing company oversight to manage compliance.