Story perspectives

Google Grants Pentagon AI Access in Classified $200 M Deal

4/28/2026

Summary

Google AI Deal

Google signed a classified deal with U.S. Department of Defense letting the Pentagon use its AI.
The amendment lets the government request changes to Google’s AI safety filters.
The contract bans AI for surveillance or weapons without human oversight and gives Google no veto.
In 2025 the Pentagon signed $200 million contracts with OpenAI, Anthropic and Google.
560 employees emailed CEO Sundar Pichai urging a block, though Reuters could not verify.

Summary

Google Workers Oppose

Hundreds of Google workers, including over 18 senior staff, signed a letter urging CEO Sundar Pichai to reject classified Pentagon AI contracts.
They argue Google cannot guarantee safeguards against lethal autonomous weapons or mass surveillance because the DoD bans technical controls.
The letter cites Anthropic’s supply-chain risk label after it refused military use, saying refusing all classified projects is the only way to avoid unethical applications.

History

Historical Context

In 2017 the Pentagon started Project Maven, a $70-million AI drone-analysis program that hired Google. In April 2018 thousands of Google engineers, urged by Sundar Pichai, signed a petition to halt the work, and Google said it would not renew the contract due in 2019. A new classified Pentagon AI deal that sidesteps Google’s safety filters revives the same ethical debate.

1 of 3

Google AI Deal

Google signed a classified deal with U.S. Department of Defense letting the Pentagon use its AI.
The amendment lets the government request changes to Google’s AI safety filters.
The contract bans AI for surveillance or weapons without human oversight and gives Google no veto.
In 2025 the Pentagon signed $200 million contracts with OpenAI, Anthropic and Google.
560 employees emailed CEO Sundar Pichai urging a block, though Reuters could not verify.

1 / 3