Pentagon announced on Friday that it has struck deals with some of the tech industry’s biggest companies in an effort to expand the military’s AI capabilities and increase the number of firms authorized to be on secret networks.
According to the Department of Defense, the companies agreed to allow the Pentagon to use their technology for “any lawful use,” a standard resisted by Anthropic, which was initially the only AI model available on classified markets.
The Pentagon has previously confirmed deals with Elon Musk’s xAI, OpenAI and Google. In addition, the Pentagon said it has made deals with Amazon Web Services, Microsoft, Nvidia and Reflection AI, a start-up.
“These agreements accelerate the transformation toward establishing the United States military as an AI-first fighting force,” the Pentagon said in a statement.
However, the announcement with the cloud computing companies is not as significant as earlier agreements with OpenAI, which allow ChatGPT and Google to use Gemini on classified networks. Cloud companies host the models, but agreements with the companies that create the models will determine exactly how the Pentagon uses AI to fight wars.
In its announcement, the Pentagon did not specify how it will use the new AI tools, but said the agreement will help service members make faster and better decisions. Some potential use cases include generating target lists to consider a strike during a conflict, such as in Iran, or analyzing large tranches of data to gather intelligence.
“Access to a diverse set of artificial intelligence capabilities from across a range of resilient American technologies will give warfighters the tools they need to act with confidence and protect the nation from any threat,” the Pentagon said.
Artificial intelligence has helped the military and spy agencies sift through vast amounts of intelligence. This was particularly valuable for finding missed communication intercepts. Large language models are also being integrated into systems like Maven that help military personnel pick out targets on the battlefield — though the Pentagon has not explained exactly how the artificial intelligence was used to find targets in Iran and elsewhere.
Defense Department officials hope the new agreements will force Anthropic to drop its reservations about the broad military standard of “any lawful use.”
Antropic did not respond to a request for comment.
President Trump has ordered the government to cut ties with Anthropic, but for now the company’s technology remains on classified networks and intelligence analysts still depend on the firm’s models. While the Pentagon wants to move quickly to OpenAI’s ChatGPT and Google’s Gemini, more and more difficulties and technical issues are emerging.
Anthropic and the Pentagon are currently locked in a federal lawsuit over the Defense Department’s decision to designate the company a supply chain risk, a novel use of government power to raise concerns about how corporations manufacture their products.
Impressed and alarmed by the power of Anthropic’s newest model, the Mythos, White House officials pushed for a compromise that would end the company’s disputes with the Pentagon, or at least allow other parts of the government to work with the firm.
While Mr Trump ordered the government to stop using Anthropic by mid-year, the success of the new model means the decision will almost certainly be reversed. The military and intelligence agencies continue to use older Anthropic models for day-to-day work and test the new model’s capabilities to find cybersecurity exploits.
The agreement with the new companies was Bloomberg reported earlier.
A Pentagon official said the new deals will help prevent “vendor lock-in” and ensure the military doesn’t have to depend on any one company. The military also wants companies to agree to a uniform standard and is reluctant to give companies contractual guarantees about how their models will be used.
Antropic and the Pentagon are debating whether Claude’s company model could be used to pilot autonomous drones or work on home surveillance. The Pentagon says it doesn’t intend to use the model for any of those activities, but the two sides haven’t agreed on contract language, or if it’s even necessary.
The details of the deals with the new companies remain confidential. But people briefed on the talks said at least some of the companies have received assurances similar to what Anthropic sought for drones and surveillance, even if the companies adhere to the overall standard the Pentagon has met.
The exact language of the guarantees could raise questions about what exactly is preventing the Pentagon and Anthropic from reaching a deal.
Pentagon and Anthropic both benefited from the fight in their own way. Anthropic has gained support in Silicon Valley for standing up to the Pentagon. And Defense Secretary Pete Hegseth was able to use the fight to criticize Anthropic’s leadership as ideological, a line Mr. Trump supports.
The Pentagon deals come as the White House resists Anthropic’s plan to expand its Mythos model to dozens of other companies, according to people familiar with the matter. The opposition stems primarily from concerns that expanding access to Mythos — which has already been shared with select organizations in the United States and Britain — could pose significant cybersecurity risks.
Mythos is so adept at finding and exploiting computer bugs that Anthropic didn’t make them public, choosing instead to give some security researchers early access to work on fixes for the vulnerabilities Mythos identified.
Asked about the Mythos discussions, a White House official said the Trump administration is trying to find a balance between advancing innovation and ensuring security as it works with leading AI labs.
Dustin Volz contributed reporting.
