Google quietly withdrew from a $100 million Pentagon drone swarm competition just weeks after qualifying, a move that internally had nothing to do with resource constraints — despite what the company told the government. The decision came after an internal ethics review, leaving multiple employees who worked on the project disappointed.
The competition, run by the Special Operations Command's Defense Autonomy Combat Operations group and the Defense Innovation Unit (the Pentagon's tech outpost), aimed to let commanders control drone swarms with voice commands. The six-month contest includes later phases focused on "target-related perception and sharing" and "from launch to termination." OpenAI, Palantir, and xAI have all been selected to continue. Anthropic also submitted a proposal but wasn't selected; the company determined it didn't cross the line CEO Dario Amodei drew on fully autonomous weapons.
Google's contradictory posture came into full view on the same day: while withdrawing from the drone swarm contest, the company amended an existing contract with the Pentagon to allow the military to use its AI for "all lawful government purposes," including classified work. The move comes as Google continues to push deeper into the AI arms race, with Cloud CEO Thomas Kurian recently teasing a new Gemini model arriving "very, very soon." A Google spokesperson said the contract change was just a routine enterprise customer adjustment for a consumer-grade security filter, "not involving development of a custom model for the Pentagon," and emphasized that "AI should not be used for autonomous weapons without appropriate human oversight."