Sam Altman navigates Anthropic's Pentagon fight as OpenAI pursues its own deal with the military

• OpenAI CEO Sam Altman has carefully weighed in on the fight between Anthropic and the Pentagon.
• He said AI companies need to work with the military, but doesn't think they should be compelled.
• Anthropic says it won't budge on two "red lines" — mass surveillance and fully autonomous weapons.

OpenAI CEO Sam Altman gingerly waded into the conflict between Anthropic and the Pentagon on Friday.

"The government, the Pentagon, needs AI models. They need AI partners," Altman, who is pursuing his own deal with the military, said on CNBC on Friday. "This is clear, and I think Anthropic and others have said they understand that as well."

Anthropic is staring down a deadline to strike a deal with the Defense Department to use its frontier model, Claude. Anthropic CEO Dario Amodei, however, said Thursday that he wouldn't budge on what he called two "red lines."

In a memo posted to Anthropic's website, Amodei said he "cannot in good conscience accede to their request" to use Claude for mass domestic surveillance and fully autonomous weapons.

The Pentagon had earlier given Anthropic an ultimatum to get on board or face a potential blacklist from government contracts, which would be a significant blow to Anthropic's bottom line.

A senior Pentagon official told Business Insider on Thursday that Defense Secretary Pete Hegseth is prepared to compel Anthropic to work with the Pentagon under the Defense Production Act, a 1950s law.

Altman said that, from his perspective, that might be a step too far, but that it's essential for AI companies to work with the government and the military.

"I don't personally think the Pentagon should be threatening DPA against these companies," he told CNBC. "But I also think that companies that choose to work with the Pentagon, as long as it is going to comply with legal protections and the few red lines that we have in the field, I think it is important to do that."

"For all the differences I have with Anthropic, I mostly trust them as a company, and I think they really do care about safety," Altman added.

OpenAI, Anthropic, xAI, and others are all competing to be the government's chosen model. While Anthropic, OpenAI, and Google have all been cleared to work with government information, only xAI's Grok has so far been cleared to handle classified information at the Pentagon.

The Wall Street Journal reported Thursday that in a note to staff, Altman said OpenAI is pursuing its own deal with the Defense Department that "allows our models to be deployed in classified environments and that fits with our principles."

He added that the effort was meant to "help de-escalate things," an apparent reference to the heated exchanges between Anthropic and the Defense Department.

OpenAI's views on working with the military have evolved in recent years. It deleted a clause from its ChatGPT "usage policies" page in 2024 that had barred "activity that has a high risk of physical harm," including "weapons development" and "military and warfare," allowing the company to pursue military contracts. OpenAI also appointed former National Security Agency Director Paul Nakasone to its board of directors in 2024.

OpenAI, Anthropic, and the Defense Department did not immediately respond to Business Insider's request for a comment.