# Dual-Use AI The concept that AI technologies developed for commercial purposes can be repurposed for military and intelligence applications — and vice versa. This is the central ethical tension in defense tech. ## The Landscape - **Google** famously pulled out of [[Project Maven]] (Pentagon AI drone targeting) in 2018 after employee protests. This created the market opening that [[Palantir]], [[Anduril]], and [[Scale AI]] filled. - **[[Anthropic]]** partnered with [[Palantir]] and Amazon to sell Claude AI models to federal intelligence and defense operations — then faced Trump administration blacklisting as a "supply chain risk" for maintaining military use restrictions. - **[[Scale AI]]** provides the data infrastructure for training and testing military AI through the Thunderforge program. - **[[Palmer Luckey]]** has explicitly rejected the "dual-use" euphemism, arguing defense tech companies should openly state they're building weapons for democracies. ## Key Tensions 1. Companies that refuse military work (Google) lose contracts to companies that embrace it (Palantir, Anduril). 2. Companies that engage with the military but set ethical boundaries (Anthropic) face political retaliation. 3. The same AI that powers chatbots and search can power autonomous weapons targeting and surveillance. 4. Domestic deployment of military tech ([[Skydio]] drones for ICE) blurs the line between national defense and domestic surveillance. ## See Also - [[Autonomous Weapons Debate]] - [[Palantir]] - [[Scale AI]] - [[PayPal Mafia to Pentagon Pipeline]]