OpenAI has secured a landmark agreement with the Pentagon to supply artificial intelligence to classified US military networks, stepping into a power vacuum created after the Trump administration abruptly cut ties with rival AI company Anthropic. The move signals a major shift in how the federal government intends to engage with the AI industry going forward.
The fallout began when Anthropic, the maker of the Claude AI system, refused to drop two key ethical restrictions from its terms of service — prohibitions on autonomous weapons capable of killing without human oversight, and restrictions on mass domestic surveillance. Pentagon officials pushed back hard, demanding unrestricted access to Claude’s capabilities in the name of national security.
President Trump personally weighed in, posting on his Truth Social platform that Anthropic had made a “disastrous mistake” by trying to “strong-arm” the Department of Defense. He directed all federal agencies to immediately cease using Anthropic technology, effectively banning the company from government contracts.
OpenAI CEO Sam Altman moved quickly to capitalize on the moment, announcing a deal with the Pentagon that he claimed includes the same ethical assurances Anthropic had sought. Altman insisted that his company’s agreement prohibits use of AI for domestic mass surveillance and autonomous lethal weapons systems, and that the Pentagon has formally agreed to these principles.
Anthropic, for its part, issued a defiant statement saying that no amount of political pressure or punishment would change its position on mass surveillance or autonomous weapons. The company maintains that these two restrictions have not impacted a single government mission to date, and says it has acted in good faith throughout all negotiations.