Pentagon Signs AI Deals with Eight Tech Companies Including Nvidia, Microsoft, and AWS for Classified Networks

The U.S. Department of Defense has signed agreements with eight technology companies including Nvidia, Microsoft, AWS, and Google to deploy AI models on classified military networks as part of building an 'AI-first fighting force,' while Anthropic remains notably absent after refusing to ease safeguards on military surveillance and autonomous weapons use. Google withdrew from a $100 million Pentagon drone-swarm competition in mid-February following internal ethics reviews and pressure from hundreds of AI researchers concerned about military applications, echoing 2018's Project Maven backlash that prompted resignations and thousands of employee signatures. The drone-swarm program, jointly led by US Special Operations Command and the Defense Innovation Unit, aims to enable voice-controlled coordination of multiple drones with capabilities including target identification and mission execution, with OpenAI, Palantir, and xAI selected as participants. The tensions reflect broader conflicts between defense agencies and tech firms over AI ethics, with a federal judge questioning whether the Pentagon's designation of Anthropic as a supply-chain risk was punitive retaliation for the company's refusal to compromise safety standards.

Google's Withdrawal from Pentagon Drone Program

• ▪Google informed Washington of its decision to withdraw from the drone-swarm program in mid-February
• ▪The US Department of Defense drone-swarm technology program is valued at up to $100 million

Technical Scope and Participants of the Defense Initiative

• ▪The Pentagon drone-swarm initiative is jointly led by the US Special Operations Command and the Defense Innovation Unit
• ▪The Pentagon drone-swarm initiative is structured as a multi-stage effort over roughly six months
• ▪Palantir Technologies was selected for the Pentagon drone-swarm program

Broader Tech Industry Tensions with Military AI Applications

• ▪In March, a US federal judge questioned whether the Pentagon's decision to designate Anthropic as a supply-chain risk was punitive
• ▪The Pentagon's designation of Anthropic as a supply-chain risk followed Anthropic's refusal to ease safeguards related to the use of Anthropic's models in military surveillance and autonomous weapons

Google's History of Internal Opposition to Defense Contracts

• ▪Google employees working on AI raised concerns about the potential use of Gemini AI in large-scale surveillance and other sensitive defense applications without meaningful human oversight
• ▪The Pentagon's Project Maven involved applying image recognition to drone footage to detect and track objects
• ▪More than 100 Google employees working on AI called on leadership to define clear limits on how Google's systems could be used in military contexts
• ▪In 2024, Google dismissed dozens of employees who protested a contract linked to Israel
• ▪In 2018, the Pentagon's Project Maven prompted internal backlash at Google, leading to resignations among senior staff and a petition signed by thousands of employees
• ▪The letter from Google AI employees was addressed to senior leadership at Google DeepMind