Pentagon Expands Google Gemini AI Use After Dropping Anthropic Over Supply Chain Risk

Pentagon AI chief Cameron Stanley confirmed to CNBC that the Department of Defense is expanding its use of Google's Gemini artificial intelligence model, about two months after the DOD dropped Anthropic, designating it as a supply chain risk.

The DOD is using Google's latest model for classified projects, according to a person with knowledge of the matter who asked not to be named because the specifics of the arrangement aren't public. The Information earlier reported that Google had signed a deal with the DOD for classified work, citing a person familiar with the matter.

In addition to Gemini, the Pentagon is also working with OpenAI and other vendors to modernize wartime capabilities, Stanley told CNBC in a video interview. "Overreliance on one vendor is never a good thing," he said. "We're seeing that, especially in software."

The DOD's embrace of Google comes amid a heated legal dispute with Anthropic. Earlier this month, a federal appeals court in Washington, D.C., denied Anthropic's request to temporarily block the department's blacklisting of the AI company as a lawsuit challenging the sanction plays out.

That ruling came after a judge in San Francisco, in a separate but related case, granted Anthropic a preliminary injunction that bars the Trump administration from enforcing a ban on the use of its Claude model. With the split decisions by the two courts, Anthropic is excluded from DOD contracts but is able to continue working with other government agencies during the litigation.

A spokesperson for the DOD confirmed over email that the agency is not working with Anthropic at this time. President Donald Trump told CNBC last week that "it's possible" there will be a deal allowing Anthropic's models to be used within the DOD.

Stanley said that by using Gemini, the Pentagon and U.S. warfighters are saving time and money. "There's a lot of different things that are saving thousands of man hours, literally thousands of man hours on a weekly basis," he said.

The arrangement is facing some opposition internally at Google, where more than 700 employees signed a letter that was sent to Google CEO Sundar Pichai this week, calling for the company to reject classified workloads. They said in the letter they don't want the technology to be "used in inhumane or extremely harmful ways."

The overarching goal, according to Stanley, is to achieve the best outcome for America's warfighters. To get there, the Pentagon has to make sure it's properly using AI models. "I have a personal quote that I usually say in these moments, you don't cook a Thanksgiving turkey in the microwave," he said. "You need to have the right technology for the right use case to achieve the right outcome."

Stanley said Anthropic's Mythos rollout earlier this month was a wakeup call. The powerful model was made available to a limited number of companies, due in part to its advanced cyber capabilities and the potential risks they posed. Stanley said the DOD is "taking this very seriously" so that it can "make sure we are not only matching the moment but are prepared for what comes next, which is a whole raft of AI-enabled capabilities" in areas that pose a challenge.