Alphabet’s Google is adding artificial intelligence (AI) agents to the US Department of Defense (Pentagon) workforce. This is for automating routine jobs. The Pentagon has three million employees. This information comes from a high-level defense official.
Emil Michael, the Under Secretary of Defense for Research and Engineering, said Google’s AI agents are called Gemini. These agents can do tasks by themselves. They will first work on unclassified networks. This will help users complete tasks they assign.
Michael explained in an interview that they are starting with unclassified networks. Most users work there. Then, they will move to classified and top-secret networks. Bloomberg reported this. Arabic Business was able to see the report.
He also said talks are happening with Google. They want to use AI agents on classified clouds. Michael is confident they will be a great partner on all networks. Jim Kelly, Google's vice president, said this too. He wrote about it on Tuesday. He said this new feature will help civilian and military staff. They can create AI agents using normal language.
Controversial Collaboration
The use of AI in the US military is controversial. Many leading US tech companies developing AI are discussing this. During the conflict with Iran, the US used AI. It helped identify targets and speed up operations. This allowed for an unprecedented intense bombing campaign.
The Pentagon is now using more products from companies like OpenAI and Google. They want to speed up AI adoption in the military field. This has caused issues with the AI company Anthropic. Anthropic wanted rules for using its tech. They wanted to prevent its use in internal monitoring and fully autonomous weapons. Because of this, the Pentagon labeled Anthropic as a supply chain risk last week.
Anthropic is suing the government. The government usually gives this label to companies from countries seen as enemies. Michael led the talks with Anthropic. He said the issue will not be solved in courts. The Pentagon is moving forward.
Google has faced internal criticism before. This was about its work with the Defense Department. In 2018, thousands of Google employees protested. They were against the company’s involvement in Project Maven. This Pentagon project used AI to analyze video from US drone strikes abroad. The backlash caused Google not to renew its contract for the program.
Later, Google removed some limits on working with the military. Emil Michael said the company is a trusted and supportive partner.
Agreement Details
At first, Google will offer 8 ready-to-use agents. Some will focus on automating tasks. These include summarizing meeting notes, preparing budgets, and checking proposed actions against national defense strategy principles. Michael mentioned this.
Some AI agents on the unclassified network may have a practical impact. They can help with planning and resource estimation for military missions and operations. A Pentagon spokesperson shared this in briefing notes.
Users can also create their own digital assistants. These will automate multi-step, repetitive administrative tasks. They do not need any programming knowledge. This is according to Google’s blog post.
Michael said training, guidance, and policies can help reduce potential risks. These risks come with using AI in the workforce. Examples include amplifying or hiding errors.
About 1.2 million Defense Department employees used Google's chatbot. They used it via the GenAI.mil portal for unclassified work. This started in December. The Gemini agents will be available starting Tuesday. However, training is very important to prevent errors. Michael stressed this. But training is far behind actual use. Only 26,000 people got AI training since December. All future training sessions organized by the Defense Department are full. This is according to the spokesperson.