Pentagon flamed after photo reveals AI request details and commander’s name

The Pentagon released a photo on social media that shows off the military's deployment of artificial intelligence (AI). But the photo shows a soldier asking it to help "create a war."

Newsweek, and others, noticed the screen on a July 21 photo from the Department of Defense Chief Technology Office bragging about its recent deployment of "GenAI Task Force," which is now being used in Hawaii at Joint Base Pearl Harbor-Hickam. The prompt read: "I NEED JHELP BUILDING AN AGENT TO CREATE A WAR I RECEIVE ," the screen capture shows. It also has a welcome at the top reading "Hello, Brandon Mizuhara, Let's get some work done!" Mizuhara is a Lieutenant Commander and is assigned to U.S. Pacific Fleet, though social media details indicate that he's serving in the Navy Medicine Readiness and Training Command.

The AI prompt is part of a Google Gemini interface.

The post on X claimed that their team crafted "more than 20 custom AI agents for sailors and helped automate tasks including operational reporting and briefing preparation," cited Newsweek.

It continues, saying the military endeavors to "systematically remove barriers to AI adoption to ensure a more lethal Department" of Defense.

Users on X were quick to track down Mizuhara and ask questions about the request on the screen. Since the question was in all caps, there was a question about what "WAR" meant in the prompt. An "agent to create a war" could mean exactly what it says, or WAR could be an acronym for "weekly activity report," used in the military and workplace. Newsweek asked the Pentagon to clarify, but never heard an answer.

The Pentagon's social media post claimed that their overall goal is to integrate AI into operations where warfighters can make faster decisions by using the tool. The downside of has been AI tools being used without human checks. In February 2026, the Trump administration bombed an elementary school, killing 120 children. The building was previously used as a military facility, but had been changed to a school ten years prior to the bombing, TIME reported in March. Palantir's Maven Smart System, an artificial intelligence operation, picked the targets in conjunction with a system by Claude.

According to Reuters, there may not have been any review by a human of the target sites before the bombing began. The report explained that the military has at its fingertips the best intelligence in the world that comes from satellites and individuals on the ground. Critics noted that a Google Street View search made it clear that the building was a school.

The bombing has raised questions about whether there should be a greater human presence in future planning.

In January, the DOD announced its AI program to "expand the military," the report said. "One of those projects, called Agent Network, is described by the department as an effort to 'unleash' AI agent development and experimentation for battle management and decision support, ranging from campaign planning to 'kill chain execution.' Another, GenAI.mil, is intended to provide department-wide access to frontier generative AI models, including Google's Gemini and xAI's Grok, for personnel at certain classification levels."

Grok was created by close Trump ally Elon Musk, who donated hundreds of millions to help get Trump elected in 2024, the Washington Post reported in Jan. 2025. Google as a corporation donated $1 million to Trump's inaugural committee.