U.S. Cyber Command is developing a flexible AI infrastructure designed to test and deploy models from any source—including potential open-source Chinese options—as it seeks to sidestep political tensions between the Pentagon and frontier AI companies. Brig. Gen. Reid Novotny, the command's chief AI officer, told Axios the goal is to create an agile system with no political constraints, adding that operators are currently "very well set for what they need right now."
The approach directly addresses the ongoing friction between Anthropic and the Department of Defense. Anthropic has restricted access to its latest Mythos Preview model due to its hacking capabilities, leaving only a patchwork of agencies—including the NSA and the Commerce Department's AI testing institute—with access. Cyber Command is effectively insulating itself from this debate by building model-swapping capabilities.
"To survive anywhere, just in case our operators want an open-source made-in-China model or something very boutique, we have to create the infrastructure and that ability to be agile—no politics," Novotny stated. The White House continues to negotiate broader access to Anthropic's models, but no resolution has been announced.
The infrastructure-first strategy positions Cyber Command to rapidly adopt top-tier AI regardless of vendor disputes or geopolitical considerations. Meanwhile, OpenAI is reportedly capitalizing on the confusion, though details of its inroads remain unclear.
Critics may argue that granting operators access to open-source Chinese models introduces unacceptable supply chain and espionage risks, even within a carefully controlled military environment.