Over 600 Google Employees Urge Sundar Pichai to Reject Pentagon Military AI Deal

More than 600 Google employees, including over 20 directors, senior directors, and vice presidents from divisions like DeepMind and Cloud, have signed a letter urging CEO Sundar Pichai to reject any deal allowing the Pentagon to use the company's AI models, such as Gemini, for classified military work. The letter, sent on Monday, April 27, 2026, warns that such access could enable "inhumane or extremely harmful ways" of deploying AI, including lethal autonomous weapons and mass surveillance, without employee oversight or the ability to intervene. As reported by Bloomberg and The Washington Post, the signatories emphasize their responsibility as AI developers to prevent unethical uses, stating that rejecting classified workloads is the only way to guarantee Google avoids association with such harms.

The employee pushback comes amid ongoing negotiations between Google and the U.S. Department of Defense to deploy Google's flagship Gemini AI in classified settings, according to reporting from The Information cited across multiple outlets including The Verge and Business Insider. Organizers of the letter highlight that AI systems can centralize power and make mistakes, raising fears of unintended consequences in secretive military applications. They argue that classified deals would bypass transparency, potentially damaging Google's reputation, business, and global role if misused. CBS News notes the workers' concerns about the technology's inappropriateness for "classified workloads," echoing broader debates in the tech industry over military AI ethics.

This action revives tensions at Google over defense contracts, following the company's 2018 withdrawal from Project Maven—a Pentagon drone imaging program—after similar employee protests. Recent context includes the Pentagon's decision two months ago to remove rival AI firm Anthropic from a contract over its restrictions on military use, as detailed by The Washington Post. In contrast, OpenAI secured a deal earlier this year with safeguards prohibiting its tech for mass domestic surveillance or directing autonomous weapons. Google has reportedly proposed its own contractual language to block Gemini from domestic surveillance or uncontrolled autonomous weapons, per letter organizers quoted in Citizen Digital, but employees deem this insufficient without a full refusal of classified access.

The letter's emergence underscores growing friction between tech workers and management on AI's military applications, especially as Google rebuilds ties with the Defense Department—such as making Gemini available to all unclassified Pentagon employees in March. No immediate response has come from Google or the Pentagon, according to CBS News and other reports. Organizers have vowed to continue protests until the company sets clear, enforceable boundaries against AI weaponization.

For Google, the stakes involve balancing lucrative defense opportunities against talent retention and public trust in an era of rapid AI advancement. Employees affected span key AI labs, potentially signaling internal risks if Pichai proceeds without concessions. What happens next remains unclear: the company could issue a statement, adjust negotiations, or face escalated activism, while the Pentagon might pivot to other vendors amid its push for AI integration in national security. This episode highlights the tech sector's pivotal role in shaping AI's boundaries for warfare and surveillance.