OpenAI Takes the Pentagon Deal Anthropic Refused — But With the Same Red Lines

Sam Altman’s decision to strike a Pentagon deal that Anthropic reportedly could not secure has placed OpenAI in an awkward but powerful position: the company that benefited most from its rival’s expulsion is also, by its own account, holding the same ethical lines that caused that expulsion in the first place.
The Anthropic situation came to a head after months of unsuccessful negotiations over the terms under which its Claude AI could be used by the US military. The company’s conditions — no autonomous weapons, no mass surveillance — were presented as minimal ethical requirements consistent with basic principles of human responsibility and civil liberties.
Defense officials were unwilling to accept those conditions. When President Trump intervened with a public directive ordering all federal agencies to stop using Anthropic technology, it ended the company’s government relationships and sent a chill through the broader AI industry about the true cost of ethical commitments.
OpenAI CEO Sam Altman announced the Pentagon deal hours later, framing it as consistent with his company’s values. In an internal memo he described autonomous weapons and mass surveillance as OpenAI’s “main red lines” — language that was all but identical to Anthropic’s stated principles. He also suggested the Pentagon should offer these terms to all AI companies.
The gap between what Anthropic was offered and what OpenAI claims to have received remains unclear, and the hundreds of AI workers who signed letters in support of Anthropic are watching closely. Whether OpenAI’s deal truly holds those ethical lines — or whether the company has simply repackaged the same concessions in more politically palatable language — will become the defining question of the partnership.