OpenAI has reached an agreement with the Department of War (DoW) to deploy its AI models within the department’s classified network, CEO Sam Altman said.
“Tonight, we reached an agreement with the Department of War to deploy our models in their classified network,” Altman wrote in a post on X.
This comes in the backdrop of the US President Donald Trump administration cracking down on Anthropic. The government cited concerns over national security and military authority after the AI company declined Pentagon’s demand for unfettered access to its Claude AI model.
Under the agreement, OpenAI said its models will operate under specific safety principles, including prohibitions on domestic mass surveillance and maintaining human responsibility for the use of force, including in autonomous weapon systems.
“AI safety and wide distribution of benefits are the core of our mission,” Altman said. “Two of our most important safety principles are prohibitions on domestic mass surveillance and human responsibility for the use of force, including for autonomous weapon systems. The DoW agrees with these principles, reflects them in law and policy, and we put them into our agreement.”
The company said it will implement technical safeguards to ensure its models operate within agreed parameters. It will also deploy field deployment engineers (FDEs) to support implementation and safety oversight. The models will run on cloud networks only.
“We also will build technical safeguards to ensure our models behave as they should, which the DoW also wanted,” Altman said. “We will deploy FDEs to help with our models and to ensure their safety, we will deploy on cloud networks only.”
Altman said OpenAI has asked the Department of War to extend similar terms to other AI companies.
“We are asking the DoW to offer these same terms to all AI companies, which, in our opinion we think everyone should be willing to accept,” he said.
The company also called for reduced legal and governmental conflict around AI partnerships.
“We have expressed our strong desire to see things de-escalate away from legal and governmental actions and towards reasonable agreements,” Altman said.
Altman added that OpenAI remains committed to broad public benefit. “We remain committed to serve all of humanity as best we can,” he said. “The world is a complicated, messy, and sometimes dangerous place.”
On February 27, the US government ordered federal agencies to cease use of Anthropic’s technology.
In a post on Truth Social, Trump said the United States “will never allow a radical left, woke company to dictate how our great military fights and wins wars,” adding that such decisions belong to the commander-in-chief and appointed military leaders. He accused Anthropic of attempting to “strong-arm the Department of War” and said its actions were “putting American lives at risk.”
Trump said there would be a six-month phase-out period for agencies currently using Anthropic’s products, including the United States Department of War. He warned that if the company did not cooperate, he would “use the Full Power of the Presidency to make them comply, with major civil and criminal consequences to follow.”
Defence Secretary Pete Hegseth said he had directed the United States Department of Defence to designate Anthropic a supply-chain risk to national security, citing the company’s refusal to comply with government demands regarding how its technology may be used.
Anthropic said it would challenge any such designation in court. “No amount of intimidation or punishment from the Department of War will change our position on mass domestic surveillance or fully autonomous weapons,” the company said.
It added that negotiations had stalled over “the mass domestic surveillance of Americans and fully autonomous weapons.”
Anthropic signed a $200 million contract with the United States Department of Defence in July last year and said it had sought assurances that its AI models would not be used for fully autonomous weapons or mass domestic surveillance. The Pentagon rejected that request and set a deadline for the company to agree to broader lawful-use terms.
ALSO READ: The Playground is Closed: 10 Hard Truths from the Cisco AI Summit
