US Military Used Anthropic AI for Iran Strikes
The U.S. military reportedly used Anthropic's Claude AI, integrated with its Maven system, to help target over 1,000 strikes in Iran in a single 24-hour period. Analysts note this marks a new era of battlefield compression, where AI's ability to process satellite and drone intel at speed is becoming more critical than troop numbers.
Project Maven, officially the Algorithmic Warfare Cross-Functional Team, was launched by the Pentagon in April 2017. Its creation was driven by concerns that the U.S. was falling behind China's military AI advancements, with an initial focus on using computer vision to analyze the overwhelming volume of video from surveillance drones. Now under the National Geospatial-Intelligence Agency, Maven's scope has expanded far beyond video analysis. The system fuses data from numerous sources, including satellites and signals intelligence, with the explicit goal of creating "fast, precise and resilient kill chains" to give warfighters "decision superiority". Anthropic was founded in 2021 by former OpenAI executives as a public benefit corporation with a stated mission of building safe and steerable AI. The company, which has received billions in funding from Google and Amazon, pioneered a training technique called "Constitutional AI" to align its models with human values, a philosophy that makes its military use particularly noteworthy. Anthropic's Claude AI was integrated into Maven through a partnership with established defense contractor Palantir Technologies. This follows a broader trend, with the Department of Defense awarding contracts to Anthropic, Google, OpenAI, and xAI in July 2025 to accelerate the use of commercial AI for military applications. The use of Claude in the Iran strikes represents a new milestone in military AI. The model reportedly synthesized intelligence to generate target packages with specific GPS coordinates, recommended weapons, and even automated legal justifications for the strikes. This moves generative AI from a background analysis tool to an active participant in targeting cycles. This deployment aligns with the Pentagon's "AI-first
