OpenAI Sidles into Pentagon as Trump Cans Anthropic

Image generated by OpenAI’s ChatGPT

Anthropic is being phased out of Pentagon systems after a dispute with the Department of War over AI safeguards, with OpenAI stepping in under a new agreement to deploy its models in classified networks.

U.S. President Donald Trump ordered federal agencies to stop using Anthropic’s technology, saying the government would “IMMEDIATELY CEASE all use of Anthropic’s technology” and announcing a “Six Month phase out period” for agencies including the Department of War. In the same statement, Trump criticized the company in political terms, saying, “The Leftwing nut jobs at Anthropic have made a DISASTROUS MISTAKE trying to STRONG-ARM the Department of War.”

U.S. Secretary of War Pete Hegseth said the Pentagon’s position “has never wavered” and that it “must have full, unrestricted access” to AI systems for “every LAWFUL purpose in defense of the Republic.” He accused Anthropic of seeking to “seize veto power over the operational decisions of the United States military” and said the company would be designated a “Supply-Chain Risk to National Security.”

Under Hegseth’s directive, no contractor, supplier or partner doing business with the U.S. military may conduct commercial activity with Anthropic. The company will continue to provide services for up to six months to allow for a transition to another provider.

The move follows months of negotiations between the Pentagon and Anthropic over two specific exceptions the company sought to maintain in its contracts governing the use of its Claude AI model.

Anthropic CEO Dario Amodei said talks reached an impasse over “two exceptions we requested to the lawful use of our AI model, Claude: the mass domestic surveillance of Americans and fully autonomous weapons.” He said the company supports “all lawful uses of AI for national security aside from the two narrow exceptions above.”

Amodei argued that current frontier AI models are “not reliable enough to be used in fully autonomous weapons” and warned that allowing such use “would endanger America’s warfighters and civilians.” On surveillance, he said “mass domestic surveillance of Americans constitutes a violation of fundamental rights.”

He also pushed back against the supply-chain designation, calling it “an unprecedented action—one historically reserved for US adversaries,” and said Anthropic would “challenge any supply chain risk designation in court.”

In a separate statement issued earlier, Amodei said the Department had insisted contractors accept “any lawful use” and remove safeguards related to surveillance and autonomous weapons. He said the company “cannot in good conscience accede to their request,” but added that if the Department chose to off-board Anthropic, it would work to “enable a smooth transition to another provider.”

Anthropic had been among the first frontier AI firms to deploy models in U.S. government classified networks and at national laboratories. According to Amodei, Claude has been used for “intelligence analysis, modeling and simulation, operational planning, cyber operations, and more.”

Within hours of the Pentagon’s decision, OpenAI CEO Sam Altman announced a new agreement with the Department of War.

“Tonight, we reached an agreement with the Department of War to deploy our models in their classified network,” Altman said. He added that “the DoW agrees” with OpenAI’s safety principles, including “prohibitions on domestic mass surveillance and human responsibility for the use of force.”

Altman said OpenAI would “build technical safeguards” and deploy systems only on “cloud networks,” and called on the Department to offer “these same terms to all AI companies.”

The shift marks a significant change in the Pentagon’s AI partnerships. As Anthropic prepares to exit under a six-month phase-out and considers legal action, OpenAI is positioned to take on a larger role in supplying AI capabilities to U.S. defense and intelligence systems. The dispute has also highlighted broader tensions over how artificial intelligence should be governed when deployed in national security operations.

Both, Anthropic and OpenAI, have received work from the Pentagon worth hundreds of millions of dollars, amounts that are expected to expediently experience exponential explosion.