OpenAI CEO Sam Altman has admitted the company’s hastily arranged deal to supply AI to the US Department of War looked “opportunistic and sloppy” and is now walking it back.

The deal came almost immediately after the Pentagon dropped its previous AI contractor, Anthropic, following a public spat with President Trump, who called the company “leftwing nut jobs” for refusing to allow its technology to be used for domestic mass surveillance.

OpenAI stepped in to fill the gap, but the move sparked fierce criticism. Nearly 900 employees from OpenAI and Google signed an open letter urging their companies to refuse DoW requests to use AI for surveillance or autonomous killing. A “delete ChatGPT” campaign spread across social media and notably, Anthropic’s Claude shot to the top of Apple’s App Store, surpassing ChatGPT in downloads.

Altman has since said OpenAI will explicitly bar its tech from mass surveillance use and from deployment by intelligence agencies like the NSA. Critics remain skeptical, with OpenAI’s former head of policy research Miles Brundage suggesting the company “caved and framed it as not caving.”

Meanwhile, three more US cabinet agencies- State, Treasury, and Health have moved to drop Anthropic’s products following Trump’s directive to phase them out across government.