OpenAI Reaches Pentagon Deal on Classified AI Use as Trump Orders Government to Drop Anthropic
The facts
- OpenAI has reached an agreement with the U.S. Department of Defense (referred to by Altman as the Department of War) to deploy its AI models on the Pentagon’s classified network.
- Sam Altman publicly announced the Pentagon agreement in a post on X, writing that OpenAI had reached a deal to deploy its models in the department’s classified network and praising the department’s respect for safety.
- Altman says two core safety principles incorporated into the Pentagon agreement are prohibitions on domestic mass surveillance and a requirement for human responsibility in decisions involving the use of force, including for autonomous weapon systems.
- OpenAI says it will build technical safeguards or controls into the Pentagon deployment to ensure its models behave within agreed safety boundaries.
- The OpenAI–Pentagon deal was announced shortly after President Donald Trump directed U.S. government agencies to immediately stop using Anthropic’s AI technology.
- Trump’s order and the Pentagon’s break with Anthropic followed a dispute in which the Defense Department sought permission to use Anthropic’s models for all lawful military purposes, which Anthropic resisted.
- Anthropic has opposed uses of its AI for mass surveillance of U.S. citizens and fully autonomous weapons, positions that contributed to its standoff with the Pentagon.
- Anthropic has indicated it is prepared to challenge the government’s actions, including describing the pressure as intimidation and vowing to sue or otherwise contest the move.
Context
What exactly does the OpenAI–Pentagon agreement cover?
According to Sam Altman, the agreement allows OpenAI’s AI models to be deployed on the Defense Department’s classified network, with technical safeguards and contractual terms that bar domestic mass surveillance and require human responsibility for decisions involving the use of force, including in autonomous weapons contexts. POLITICO,TimesNow,NTD,Hindustan Times,Hindustan Times,Indian Express,MoneyControl,MoneyControl,India Today,WSJ,Al Jazeera Online
Why did the U.S. government move away from Anthropic’s AI technology?
The Pentagon sought the ability to use Anthropic’s models for all lawful military purposes, including areas like surveillance and autonomous weapons, but Anthropic refused to lift its restrictions on uses such as mass surveillance of U.S. citizens and fully autonomous weapons, prompting President Trump to order agencies to stop using Anthropic’s technology and the Defense Department to sever ties. CNBC,Hindustan Times,Hindustan Times,NY Post,Newsweek,Yahoo! Finance,WSJ,Business Insider
How does OpenAI’s stance on military AI use compare to Anthropic’s?
Both companies have articulated red lines against mass surveillance and fully autonomous lethal weapons, but while Anthropic refused to modify its terms to allow unrestricted military use, OpenAI says it has reached a deal that preserves prohibitions on domestic mass surveillance and autonomous offensive weapons while still enabling its models to operate in classified military environments. TimesNow,NTD,CNBC,Hindustan Times,Hindustan Times,NY Post,India Today,WSJ,Business Insider,Al Jazeera Online
The receipts — 100 sources
Wire services (6)
Other coverage (94)
A news briefing built to show both sides.
The day’s biggest stories in one short brief, built to show both sides and the competing values behind the headlines. Free in your inbox.