Anthropic Pitched Swarm Tech to Pentagon
AI company Anthropic pitched voice-controlled, autonomous drone swarm technology as part of the Pentagon’s new $100 million contest. The proposal aims to develop robust and scalable swarms for defense applications, highlighting the growing demand for AI in multi-agent systems.
Anthropic's proposal was submitted to a six-month competition run by the Pentagon's Defense Innovation Unit (DIU) and the newly formed Defense Autonomous Warfare Group. The challenge offers a $100 million prize for a system that allows a single operator to command autonomous drone swarms using voice commands in plain English. This initiative is part of a wider AI acceleration strategy to enhance military planning, logistics, and combat systems. The pitch occurred amidst a significant dispute between Anthropic and the Department of Defense (DoD) over ethical red lines. Anthropic has insisted on safeguards preventing its AI from being used for mass surveillance of American citizens or in fully autonomous weapons that can engage targets without human intervention. This stance has led to tensions, with the Pentagon reportedly threatening to label the company a "supply chain risk." The competition has attracted other major players, including SpaceX and its subsidiary xAI, positioning them as direct competitors in this high-stakes challenge. OpenAI is also involved, though in a supporting role, by providing technology to translate a commander's verbal orders into digital instructions for two other defense tech firms. This push for AI-driven swarms is a direct response to the evolving nature of warfare, heavily influenced by the widespread use of inexpensive drones in recent conflicts like the one in Ukraine. The goal is to overcome the limitations of current drone light shows, which rely on pre-programmed routes, and develop resilient, decentralized swarms that can operate under hostile conditions like electronic attacks. The program is a successor in spirit to initiatives like Project Maven, a DoD effort started in 2017 to accelerate the adoption of machine learning for analyzing surveillance data. That project, initially involving Google, also faced ethical backlash from employees, leading to Google's withdrawal. Now, contractors for Maven include Palantir, Anduril, and Amazon Web Services. The Pentagon's AI efforts are now spearheaded by the Chief Digital and Artificial Intelligence Office (CDAO), which was formed in 2022 by merging the Joint Artificial Intelligence Center (JAIC) and other digital services units. The CDAO's mission is to accelerate the delivery and adoption of AI at scale across the entire DoD.
