Pentagon AI chief confirms expanded use of Google by Department of Defense

The Pentagon's AI leader, Cameron Stanley has shared that the US Department of Defense is increasing its use of Google's Gemini AI technology.

This decision comes approximately two months after the DOD labeled Anthropic as a risk to the supply chain and ceased its partnership.

According to someone informed about the agreement, who preferred to remain anonymous due to confidentiality, the DOD is utilising Google's current model for covert projects.

Earlier, The Information shared the news that Google had entered into a classified work agreement with the DOD, based on a source familiar with the details.

Beyond using Gemini, Stanley mentioned in a video chat with CNBC that the Pentagon is also collaborating with OpenAI and various suppliers to update military technologies.

"Relying heavily on a single supplier is rarely wise," he expressed. "It's becoming evident, particularly with software."

The Pentagon's collaboration with Google unfolds amidst a significant legal clash with Anthropic.

Recently, a federal court in Washington, D.C., declined Anthropic's plea to temporarily prevent the department's embargo on the AI entity, as a legal dispute over the restriction is ongoing.

This decision followed a previous ruling by a San Francisco judge in a separate related lawsuit, which temporarily halted the enforcement of a prohibition against the Claude model from the Trump administration.

Due to differing judgments from the two courts, Anthropic's role in DOD contracts is paused, but it can still engage with other government bodies while the legal process continues.

A DOD spokesperson confirmed through email that the department isn't engaged with Anthropic currently.

President Donald Trump told CNBC last week that a potential deal might emerge, permitting the use of Anthropic's models in the Department of Defense.

Stanley highlighted that by implementing Gemini, both the Pentagon and American military are achieving considerable savings in effort and resources.

"It's resulting in thousands of manpower hours being saved each week," he noted.

There is internal disapproval at Google regarding this arrangement, with over 700 employees having signed a letter addressed to Google’s CEO Sundar Pichai, urging the company not to accept classified assignments.

They expressed in the letter their concerns about the technology being utilised in unethical or profoundly damaging ways.

Stanley remarked that the launch of Anthropic's Mythos earlier this month served as a wake-up call.

The advanced model was made available only to a select few companies due to its sophisticated cyber capacity and the potential threats involved.

Stanley impressed upon the seriousness the DOD is dedicating to this matter to ensure they are not just risen to the occasion but also ready for forthcoming developments, bringing forth a wide array of AI-powered capabilities in challenging areas.