The Pentagon is seen from an airplane, Monday, Feb. 2, 2026, in Washington.
Julia Demaree Nikhinson/Related Press
conceal caption
toggle caption
Julia Demaree Nikhinson/Related Press
OpenAI CEO Sam Altman says he shares the “crimson traces” set by rival Anthropic proscribing how the navy makes use of AI fashions, amid Anthropic’s escalating feud with the Pentagon.
The Division of Protection has given Anthropic a deadline of 5:01 p.m. ET right this moment to drop restrictions on its AI mannequin, Claude, from getting used for home mass surveillance or solely autonomous weapons. The Pentagon has stated it does not intend to make use of AI in these methods, however requires AI firms to permit their fashions for use “for all lawful functions.”
Protection officers say if Anthropic does not comply, it may lose its contract price as a lot as $200 million with the U.S. navy.
The federal government has additionally threatened to invoke the Korean Conflict-era Protection Manufacturing Act (DPA) to compel Anthropic to permit use of its instruments and has, on the similar time, warned it will label Anthropic a “provide chain danger,” doubtlessly blacklisting it from profitable authorities contracts.
By wading into the standoff between Anthropic and the Pentagon, Altman may complicate the Pentagon’s efforts to switch Anthropic if it follows via on its menace to cancel the contract. OpenAI additionally has a Protection Division contract, together with Google, xAI, and Anthropic, however Anthropic was the primary to be cleared to be used on categorized techniques.
“I do not personally assume the Pentagon needs to be threatening DPA in opposition to these firms,” Altman advised CNBC in an interview on Friday morning. He stated he thinks it is necessary for firms to work with the navy “so long as it’s going to adjust to authorized protections” and “the few crimson traces” that “we share with Anthropic and that different firms additionally independently agree with.”
“For all of the variations I’ve with Anthropic, I principally belief them as an organization, and I feel they actually do care about security, and I have been completely satisfied that they have been supporting our warfighters,” Altman added. “I am undecided the place that is going to go.”
In an inner be aware despatched to workers on Thursday night, Altman stated OpenAI was looking for to barter a take care of the Pentagon to deploy its fashions in categorized techniques with exclusions stopping use for surveillance within the U.S. or to energy autonomous weapons with out human approval, in line with an individual accustomed to the message who was not approved to talk publicly. The Wall Road Journal first reported Altman’s be aware to workers.
The Protection Division did not reply to a request for touch upon Altman’s statements.
Whether or not AI firms can set restrictions on how the federal government makes use of their expertise has emerged as a serious sticking level in current months between Anthropic and the Trump administration.
On Thursday, Anthropic CEO Dario Amodei stated the Pentagon’s threats over its contract wouldn’t make the corporate budge. “We can’t in good conscience accede to their request,” he wrote in a prolonged assertion.
“Anthropic understands that the Division of Conflict, not personal firms, makes navy choices. We’ve got by no means raised objections to specific navy operations nor tried to restrict use of our expertise in an advert hoc method,” he stated, utilizing the Pentagon’s rebranded “Division of Conflict” moniker. However, he added, home mass surveillance and absolutely autonomous weapons are makes use of which are “merely exterior the bounds of what right this moment’s expertise can safely and reliably do.”
Emil Michael, the Pentagon’s undersecretary for analysis and engineering, shot again in a submit on X, accusing Amodei of mendacity and having a “God-complex.”
“He desires nothing greater than to attempt to personally management the US Navy and is okay placing our nation’s security in danger,” Michael wrote. “The @DeptofWar will ALWAYS adhere to the regulation however not bend to whims of anyone for-profit tech firm,” he wrote.
In an interview with CBS Information, Michael stated federal regulation and Pentagon insurance policies already bar using AI for home mass surveillance and autonomous weapons.”
“At some degree, it’s a must to belief your navy to do the precise factor,” he stated.
Impartial specialists say the standoff is very uncommon on the planet of Pentagon contracting.
“That is totally different for positive,” stated Jerry McGinn, director of the Middle for the Industrial Base on the Middle for Strategic and Worldwide Research, a Washington DC assume tank. Pentagon contractors do not normally get to inform the Protection Division how their services and products can be utilized, he notes “as a result of in any other case you would be negotiating use circumstances for each contract, and that is not cheap to anticipate.”
On the similar time, McGinn notes, synthetic intelligence is a brand new and largely untested expertise. “This can be a very uncommon, very public struggle,” he stated. “I feel it is reflective of the character of AI.”
