Google is strengthening its collaboration with the defense sector once again, signing a new agreement with the U.S. Department of Defense. According to sources, the Pentagon has gained access to Google’s Gemini AI models for use in classified projects, and the company will have no control over how its technology is deployed.
The terms of the deal allow the military to employ the AI for any lawful government tasks. Google may suggest restrictions—for instance, against use in autonomous weapons or mass surveillance—but these recommendations are not binding. In practice, this means the final decision rests with the government.
The contract came about after the government walked away from a partnership with Anthropic, whose AI usage restrictions didn’t satisfy U.S. authorities. Now the Pentagon is betting on Google, along with OpenAI and xAI, building a new ecosystem of technology suppliers for defense needs. According to defense officials, AI use is already saving military personnel a significant amount of working time.
Inside Google, the situation is also causing tension. Hundreds of employees have opposed the company’s involvement in such projects, citing risks to society and potential misuse of the technology. Nonetheless, competition in the AI market and pressure from other players already working with the military seem to have pushed the company toward this decision.