The US Department of Defense has signed an agreement with xAI to deploy the Grok chatbot on GenAI.mil. The Pentagon said the tool will support internal, unclassified work. Defense officials said GenAI.mil aims to reach about three million military and civilian personnel. The department plans to expand the platform with several commercial AI models.
The Pentagon said it will embed xAI’s Grok model family into GenAI.mil on December 22. The department has announced an initial deployment for early 2026. The rollout will cover military and civilian users. The release mentioned that Grok will run at Impact Level 5, a security tier for defense systems. This tier supports Controlled Unclassified Information, also known as CUI.
The Pentagon also informed that the integration will add real-time insights from the X platform. Defense officials said the department will scale a secure AI ecosystem for speed and decision support. The statement described newly certified IL5 tools as daily operational assets. The department said it expects users to integrate models into routine workflows.
The Pentagon launched GenAI.mil on December 9 and started with Google’s Gemini for Government, according to official and vendor releases. The department described GenAI.mil as a bespoke platform for approved generative AI tools. It also said it will offer training to staff who use it.
Officials and industry reports listed early use cases across research, writing, and document formatting. They also cited faster analysis of video and imagery in unclassified workflows. Google said the system will not train public models on Defense Department data.
The Pentagon’s Chief Digital and Artificial Intelligence Office has also pursued a broader vendor strategy. In July, the CDAO announced contract awards to Anthropic, Google, OpenAI, and xAI, each with a $200 million ceiling. The office said it wants commercial models to support mission, intelligence, and business systems.
Grok has drawn scrutiny over public outputs in mid-2025. Some responses included antisemitic content and praise for Adolf Hitler. xAI said it removed posts and added controls to block hate speech. A Turkish court ordered restrictions on some Grok outputs, according to Reuters.
Federal procurement decisions have also kept Grok in the spotlight. Reporting in August said the White House urged the General Services Administration to restore Grok to an approved vendor list. In September, GSA announced a OneGov agreement that priced Grok access at $0.42 per agency.
The Pentagon tied the Grok rollout to IL5 rules and unclassified tasks, not classified operations. However, defense leaders have signaled interest in broader AI use over time. For now, GenAI.mil positions Grok alongside Gemini as the platform expands.
Also Read: White House Slams Brakes on NVIDIA’s AI Chip Sales to China