Pentagon AI chief confirms work with Google after Anthropic blacklist

Pentagon AI chief Cameron Stanley confirmed to CNBC that the Division of Protection is increasing its use of Google’s Gemini synthetic intelligence mannequin, about two months after the DOD dropped Anthropic, designating it as a provide chain danger.

The DOD is utilizing Google’s newest mannequin for categorized initiatives, in line with an individual with data of the matter who requested to not be named as a result of the specifics of the association aren’t public. The Data earlier reported that Google had signed a take care of the DOD for categorized work, citing an individual aware of the matter.

Along with Gemini, the Pentagon can be working with OpenAI and different distributors to modernize wartime capabilities, Stanley advised CNBC in a video interview.

“Overreliance on one vendor is rarely a very good factor,” he mentioned. “We’re seeing that, particularly in software program.”

The DOD’s embrace of Google comes amid a heated authorized dispute with Anthropic. Earlier this month, a federal appeals court docket in Washington, D.C., denied Anthropic’s request to quickly block the division’s blacklisting of the AI firm as a lawsuit difficult the sanction performs out.

That ruling got here after a decide in San Francisco, in a separate however associated case, granted Anthropic a preliminary injunction that bars the Trump administration from imposing a ban on using its Claude mannequin. With the break up choices by the 2 courts, Anthropic is excluded from DOD contracts however is ready to proceed working with different authorities businesses in the course of the litigation.

A spokesperson for the DOD confirmed over electronic mail that the company shouldn’t be working with Anthropic right now. President Donald Trump advised CNBC final week that “it is potential” there will likely be a deal permitting Anthropic’s fashions for use throughout the DOD.

Stanley mentioned that through the use of Gemini, the Pentagon and U.S. warfighters are saving money and time.

“There’s a number of various things which might be saving hundreds of man hours, actually hundreds of man hours on a weekly foundation,” he mentioned.

The association is going through some opposition internally at Google, the place greater than 700 workers signed a letter that was despatched to Google CEO Sundar Pichai this week, calling for the corporate to reject categorized workloads. They mentioned within the letter they do not need the expertise to be “utilized in inhumane or extraordinarily dangerous methods.”

The overarching objective, in line with Stanley, is to realize the very best final result for America’s warfighters. To get there, the Pentagon has to ensure it is correctly utilizing AI fashions.

“I’ve a private quote that I normally say in these moments, you do not cook dinner a Thanksgiving turkey within the microwave,” he mentioned. “You want to have the best expertise for the best use case to realize the best final result.”

Stanley mentioned Anthropic’s Mythos rollout earlier this month was a wakeup name. The highly effective mannequin was made obtainable to a restricted variety of firms, due partially to its superior cyber capabilities and the potential dangers they posed.

Stanley mentioned the DOD is “taking this very critically” in order that it could actually “ensure we aren’t solely matching the second however are ready for what comes subsequent, which is a complete raft of AI-enabled capabilities” in areas that pose a problem.

—CNBC’s Jennifer Elias contributed to this report.

WATCH: Google, Pentagon in talks to deploy Gemini in categorized techniques