Unlock the White House Watch newsletter for free
Your guide to what Trump’s second term means for Washington, business and the world
The Pentagon has struck new agreements with tech companies to use their AI tools and infrastructure on classified networks, as the US military broadens its AI options after its relationship with Anthropic collapsed.
The defence department said on Friday it has reached deals with Nvidia, Microsoft, Reflection AI and Amazon Web Services for the “lawful operational use” of their technology. The deals are similar to those previously struck with SpaceX, OpenAI and Google.
The deals come amid a clash between the Pentagon and Anthropic. Its Claude model was previously the only one used in classified settings, including in the war against Iran and the capture of Venezuelan leader Nicolás Maduro.
But Dario Amodei, Anthropic’s chief executive, has said he refused to sign a deal with the defence department unless the government guaranteed that Anthropic’s tools would not be used for mass domestic surveillance and lethal autonomous weapons.
The seven agreements signed by the Pentagon provide access to proprietary and closed models via SpaceX, OpenAI and Google, and open source models with Nvidia and Reflection AI. Microsoft and AWS are providing AI infrastructure.
Emil Michael, the Pentagon’s chief technology officer, told CNBC on Friday that “open source is a brand new vector for us, which is very important because Chinese open source models have infiltrated a lot of different companies. We want to have an American alternative to that.”
On Monday, Google signed its Pentagon deal as more than 560 employees sent an open letter to chief executive Sundar Pichai urging him to walk away from talks because of concerns that its technology could be used in “inhumane or extremely harmful ways”.
Accelerating the incorporation of AI throughout the military — including on the battlefield — has been a priority for the Pentagon under defence secretary Pete Hegseth.
“These agreements accelerate the transformation toward establishing the United States military as an AI-first fighting force and will strengthen our warfighters’ ability to maintain decision superiority across all domains of warfare,” the Pentagon said.
The US attempt to designate Anthropic as a supply chain risk in late February — just one day before the start of the US war on Iran — was an unprecedented move by the Pentagon against an American company.
“Anthropic is run by an ideological lunatic” who should not have “decision-making over what we do”, Hegseth told Congress on Thursday.
US President Donald Trump also banned Anthropic from all government contracts, while Anthropic and the Pentagon are locked in legal battles.
However, Amodei met White House chief of staff Susie Wiles last month, as the US government sought access to Anthropic’s new Mythos model, which has shown strong advances in breaking cyber defences.
Michael said on Friday that “the Mythos issue . . . is a separate national security moment where we have to make sure that our networks are hardened up, because that model has capabilities that are particular to finding cyber vulnerabilities and patching them”.