OpenAI said today it’s working with the U.S. Department of Defense on a pilot program that will see it create “frontier AI” capabilities, but it hasn’t divulged exactly what it’s building.

The DoD confirmed the arrangement in its daily list of newly-awarded contracts today, where it revealed it will award OpenAI up to $200 million for the project, with $2 million being paid up front. “Under this award, the performer will develop prototype frontier AI capabilities to address critical national security challenges in both warfighting and enterprise domains,” the DOD said in the document.

The company broke the news through a blog post, saying the contract is part of a larger initiative known as “OpenAI for Government,” which is all about making the company’s advanced artificial intelligence models available to U.S. government agencies. In the post, OpenAI says it will be working on a prototype project that aims to explore “how frontier AI can transform [the DOD’s] administrative options,” such as by helping service members to apply for healthcare and enhancing its cyber defense capabilities.

OpenAI’s blog post does not use the word “warfighting,” and also points out that any applications of its technology will be consistent with its usage policies and guidelines, which prohibit it from being used to develop or control weapons systems. Still, if it can enhance the DOD’s cyber defense capabilities, that would probably aid in terms of being able to wage war, without technically being a weapon.

The U.S. government has shown a keen interest in applying AI technology to defense and other industries in recent months. In December, the DOD signed a $100 million contract with a company called Anduril Industries Inc., which develops high-tech vehicles and other systems for the military. Weeks earlier, OpenAI’s rival Anthropic PBC said it is working with Amazon Web Services Inc. and Palantir Technologies Inc. to supply its AI models to U.S. defense and intelligence agencies.

OpenAI co-founder and Chief Executive Sam Altman told former U.S. National Security Agency chief Paul Nakasone at a public talk in April that the company is “proud to and really wants to engage in national security areas.”

“Artificial intelligence is changing how every industry goes about its work, and the military and defense industries are no different,” said analyst Holger Mueller of Constellation Research Inc. “It’s encouraging to see the readiness to experiment and evaluate how AI can be applied to defense, but the tricky questions pertain to where the data resides, how the AI model supply chains are secured, and how you control access rights for users.”

Last week, OpenAI Chief Product Officer Kevin Weil and former Chief Revenue Officer Bob McGrew were officially sworn into the U.S. Army Reserve and given the rank of lieutenant colonel. They were joined by the chief technology officers of Palantir and Meta Platforms Inc., and will work with the newly formed Detachment 201: Executive Innovation Corps, a special unit set up to advise the U.S. government on how AI can be used by the military.

OpenAI has been cultivating close ties with the government in other ways, with one of the most visible initiatives being Project Stargate, which aims to invest $500 billion in building AI computing infrastructure in the U.S.

Image: SiliconANGLE/Dreamina