April 29, 2026 Google has signed an agreement with the U.S. Department of Defense allowing its artificial intelligence models to be used in classified government work. The deal grants the Pentagon authority to apply Google’s AI for “any lawful government purpose,” placing the company directly in the expanding market for military AI systems.
The agreement comes a day after more than 600 Google employees, including senior leaders, urged CEO Sundar Pichai to block such use.
“We want to see AI benefit humanity; not to see it being used in inhumane or extremely harmful ways. This includes lethal autonomous weapons and mass surveillance but extends beyond,” the letter said.
Google said the arrangement aligns with broader industry efforts to support national security while maintaining safeguards. A company spokesperson stated it is “proud to be part of a broad consortium” supporting defence applications, citing use cases such as logistics, cybersecurity, translation, fleet maintenance and critical infrastructure protection. The company added it remains committed to limiting AI use in domestic mass surveillance and autonomous weapons without human oversight.
However, the contract language also specifies that Google does not retain authority to “control or veto lawful Government operational decision-making.” That distinction reflects a key tension in current defence partnerships: companies may impose technical safeguards, but ultimate deployment decisions sit with government agencies.
The deal places Google alongside other major AI firms already working with the Pentagon. OpenAI and xAI have signed similar agreements allowing classified use of their models, with OpenAI stating it maintains control over its “safety stack” and restricts use for mass domestic surveillance or directing lethal autonomous systems.
Earlier negotiations with Anthropic broke down after disagreements over similar “any lawful purpose” language, particularly around surveillance and weapons applications. The U.S. government subsequently cut ties and designated the company a supply chain risk, though officials have since indicated that renewed talks remain possible.
Google’s move also marks a reversal from its 2018 decision to exit Project Maven, a Pentagon initiative involving AI analysis of drone footage, following employee protests at the time. The current agreement suggests a shift in how major technology firms are approaching defence work as demand for AI capabilities accelerates.
The broader policy debate is evolving in parallel. Lawmakers have recently introduced proposals to limit how AI can be applied to data collected under Foreign Intelligence Surveillance Act Section 702, citing concerns that AI could expand the scale and speed of intelligence analysis involving communications data.
