The U.S. Department of Defense is integrating more generative AI tools into its daily office operations. On September 1st, the Pentagon announced that its internal platform GenAI.mil has added military versions of ChatGPT and Grok, serving approximately 3 million military and civilian personnel.

GenAI.mil Two new models added

GenAI.mil was launched in December last year, initially providing only a military version of Google Gemini. The Pentagon stated that the platform currently has 1.7 million users. This expansion occurred after the U.S. Department of Defense awarded contracts worth up to $200 million to OpenAI, entities that provide Grok services, as well as Google and Anthropic.

Can process non-classified and sensitive information.

The Pentagon stated that the military versions of ChatGPT and Grok can be used in controlled, non-classified information environments, applicable to Impact Level 5, which refers to sensitive government information that has not been classified as confidential. The relevant data will be kept in a closed environment and will not be used for training models.

OpenAI stated that integrating ChatGPT with GenAI.mil will be used to improve the efficiency of non-classified work, freeing up more time and resources for task execution. The report mentioned that a large portion of the military's workload is concentrated on labor-intensive tasks such as manpower management, logistics, and contract management, which are more suitable for the implementation of large models first.

Grok Emphasizes reasoning and logistics scenarios

The Pentagon's introduction to Grok focuses more on its reasoning capabilities. The announcement states that military personnel will be able to use its “deep thinking” reasoning functions and switch between different modes such as automatic, fast, and expert. As examples, the official mentioned applications including market research analysis for procurement staff and supply chain management for logistics personnel.

Claude was not included this time.

This deployment is in line with the White House's push for the military to accelerate the adoption of AI. However, the newly added list does not include Claude of Anthropic. Reports indicate that Anthropic once pressured the Pentagon during contract negotiations to commit not to use its technology for autonomous weapons or surveillance purposes. Just last week, a judge ruled that the U.S. Department of Defense's practice of listing Anthropic as a supply chain risk was illegal.