Pentagon Seeks Custom AI from OpenAI for Military Commands
The Pentagon has requested OpenAI to create an AI system with a high compliance rate for military directives. This initiative follows a recent contract termination between the Department of Defense and AI firm Anthropic over ethical concerns.
Sources
The Department of Defense requested OpenAI to develop a customized AI for military use with "minimal refusal rates" for Pentagon commands, as revealed by documents obtained through a Freedom of Information Act lawsuit. OpenAI, along with other AI companies like Google and Anthropic, agreed in 2025 to create militarized AI prototypes under contracts potentially worth up to $200 million over two years. Although both OpenAI and the Pentagon denied agreeing to the "minimal refusal" clause, a Department of Justice attorney initially confirmed it was part of the executed contract before later retracting that confirmation, leading to confusion about the document's status.
The Department of Defense (DoD) and AI firm Anthropic ended their relationship amid controversy over military ethics, with details emerging from over 400 pages of documents obtained through a Freedom of Information Act lawsuit. Contracts signed in July 2025 involved Anthropic, Google, OpenAI, and xAI, each potentially worth up to $200 million, to develop prototypes of AI tools aimed at enhancing military decision-making and capabilities across various applications. The documents reveal a deep collaboration between the Pentagon and these tech companies, emphasizing a shared commitment to utilizing AI for military purposes, including surveillance and targeting.