Openai said Monday that the U.S. Department of Defense has granted contracts of up to $200 million to help identify and build prototype systems that use frontier models for management tasks and more.
Openai offers several examples of possible tasks, such as helping service members get health care, streamlining data for various programs, and “supporting aggressive cyber defense.” The company also said, “All use cases must match Openai’s usage policy and guidelines.”
The DoD announcement used simpler language. “Under this award, performers will develop prototype frontier AI capabilities to address key national security challenges in both the combat and enterprise domains.”
It is not yet seen whether references to war warfare apply to the weapons themselves or to other areas related to war, such as documents. Openai guidelines prohibit individual users from developing or using weapons using ChatGPT or its API. However, Openai removed the explicit “military and war” ban in its Terms of Service in January 2024.
Given how strong a number of powerful people in Silicon Valley are warning about the dangers of China’s advanced LLM models, it’s no surprise that DOD wants to use Openai for any purpose. For example, Marc Andreessen, co-founder of the VC company Andreessen Horowitz, an investor at Openai, recently appeared on Jack Altman’s “no cap” podcast (a brother of Jack is Sam Altman). Andreessen described the competition between China’s AI and Western world models as the “Cold War.”
Still, perhaps an equally interesting part of this announcement is that Openai is saying about its increasingly tense relationship with leading investor Microsoft.
Microsoft has thousands of contracts with the federal government worth hundreds of millions of dollars. For decades, we have implemented the strict security protocols necessary for governments, especially DOD, to use the cloud.
Openai announced the deal as part of a broader new “Openai for Government” program that integrates many other programs used to sell directly to government agencies, including the U.S. National Laboratory, Air Force Research Institute, NASA, NIH, and the Treasury Department.
However, it was only in April that Microsoft announced that DOD had approved the Azure Openai service at all classified levels. Now the DOD is also direct to the source. From Microsoft’s perspective: It hurts.
Neither Openai nor Microsoft responded to requests for comment immediately.