OpenAI Agrees to Pentagon Terms That Anthropic Fought and Lost Over

The Trump administration’s decision to ban Anthropic from all federal contracts has resulted in OpenAI stepping in with a deal that, on paper at least, includes the very same ethical protections Anthropic was punished for demanding. The irony of the outcome has not been lost on industry observers or AI company employees.
Anthropic’s Claude system had been at the center of a months-long standoff with the Pentagon, which wanted the AI available for military use without the company’s ethical restrictions. Anthropic drew two clear lines — no use in autonomous weapons, no use in mass surveillance — and held firm even when the administration threatened consequences.
Those consequences materialized when President Trump personally ordered federal agencies to stop using Anthropic technology, accusing the company of trying to strong-arm the military. The administration’s heavy-handed response alarmed workers across the AI industry, with nearly 500 employees from OpenAI and Google jointly signing a letter of support for Anthropic.
Hours later, OpenAI CEO Sam Altman announced that his company had reached an agreement with the Pentagon that explicitly included restrictions on mass surveillance and autonomous weapons. Altman even suggested the Pentagon should offer these same terms industry-wide — echoing almost exactly the position Anthropic had taken.
Anthropic responded to the entire episode with quiet defiance, reiterating that its principles on surveillance and autonomous weapons are non-negotiable. The company noted that its ethical restrictions had never once interfered with a legitimate government mission, framing the standoff as a political dispute rather than a practical one.