Quick overview

---

OpenAI has chosen to deploy its own AI models within the Defense Department’s classified network after rival Anthropic PBC’s relationship with the Pentagon fell apart amid concerns about surveillance and autonomous weapons.

OpenAI CEO Sam Altman said late Friday that he and the department had reached a consensus that supports the company’s position against “human responsibility for the use of force, including for autonomous weapon systems, and domestic mass surveillance.”

Altman posted on the social media platform X, saying, “As part of the deployment, the startup also built safeguards to ensure its models behave as they should.” OpenAI declined to comment on whether Anthropic’s work would be replaced by the company’s services for the department.

The Defense Department did not immediately respond to a request for comment late Friday evening.

Anthropic stated on Friday that “no amount of intimidation or punishment from the Department of War will change our position on mass domestic surveillance or fully autonomous weapons.” Anthropic has stipulated that its products cannot be used for surveillance on Americans or to carry out strikes without human involvement.

The Pentagon offered Anthropic terms earlier this week that included some language that the company had proposed on surveillance and autonomy.

However, according to the source, Anthropic did not go far enough to guarantee that the department would not be able to impose any restrictions when it felt the need to do so.

The Trump administration and Anthropic, which has garnered strong support for its position in Silicon Valley, where tech workers rallied to the company’s side and urged it to, are at risk of growing more divided as a result of OpenAI’s agreement with the Pentagon.