The Pentagon is expanding the use of AI across its workforce by introducing new digital assistants built with Google’s Gemini technology. The new system will first run on the department’s unclassified networks, where most of its workforce operates. Officials say the rollout will allow them to test the tools on a large scale before deciding whether they should be used in more sensitive systems, including classified and top-secret environments. The assistants are designed to help the defence department employees complete routine office work more quickly. Authorities believe the tools can reduce time spent on administrative tasks and allow military and civilian staff to focus more on planning, analysis, and other important responsibilities across the department.
The developments follow a dispute between the Pentagon and Anthropic. The AI giant has recently sued the Trump administration after being designated as a ‘supply chain risk’. Anthropic said that the decision was unprecedented and unlawful and argued it could cause serious harm to the company. The company was labelled as a risk after Anthropic refused to allow the US government to use its technology for autonomous weapons or domestic surveillance.
Also read: Apple iPhone 16 Plus deal: Get over Rs 21,500 discount, here’s how
Furthermore, the Google AI chief, Jeff Dean, along with several key employees from Google and OpenAI, has also signed an amicus brief supporting Anthropic in its legal challenge.
Amidst this tension, the US government is doubling down on Google. According to the senior defence official Emil Michael, the Pentagon is now introducing new digital assistants powered by Google Gemini to its unclassified networks. Unlike standard chatbots, these AI agents are capable of performing tasks independently once assigned. Furthermore, using a no-code tool called Agent Designer, employees can also build these assistants using simple language.
The first rollout will target the Pentagon’s unclassified networks, where most of the roughly three million Department of Defence workers operate. It’s expected to come with eight specialised agents designed to summarise complex meeting notes, draft detailed budget outlines, review proposed military actions against the national defence strategy and more. Officials say some agents could also support planning tasks and help estimate resources needed for military operations.
Bloomberg also reported that these capabilities will be available to the personnel through the Pentagon’s enterprise AI portal, GenAI.mil. Since the portal launched in December, about 1.2 million employees have used it, generating around 40 million prompts and uploading more than four million documents.
Also read: OpenAI to soon integrate Sora AI video tool into ChatGPT: Report
Despite the high usage, training remains limited, as only about 26,000 personnel have received formal instruction on how to use the system so far. Defence officials say expanding training is essential to ensure users review AI outputs carefully and prevent errors.