The Pentagon is accelerating deployment of artificial intelligence across military operations while moving to unwind its relationship with Anthropic after a clash over usage controls, according to officials and experts. An internal memo ordered Anthropic’s tools removed from Defense Department systems within six months, even as sources say AI—potentially including Anthropic’s Claude on classified networks—has supported U.S. actions against Iran by sifting intelligence, building targeting packages and compressing strike timelines. Former and current advisers describe AI as a force multiplier that processes battlefield video, imagery and documents at speeds no human team can match, with humans still making final targeting decisions. The Pentagon canceled a $200 million integration deal with Anthropic; the company has sued the government, alleging unlawful retaliation, and drew support from Microsoft and workers at OpenAI and Google via amicus briefs. With Anthropic sidelined, rivals are moving in: Google announced non-classified AI agents for defense uses, and OpenAI signaled interest in classified deployments under self-imposed limits. Traditional contractors still supply most weaponry, but officials say AI’s role in planning and decision support is set to expand across future campaigns.
Related articles:
— Project Maven
— Lethal autonomous weapon
— Joint All-Domain Command and Control (JADC2)
— Iron Dome
— Explainable artificial intelligence
