Dramatically upturning the military’s artificial intelligence (AI) landscape, OpenAI has reached an agreement to deploy its models within the Department of Defense’s classified networks – at the expense of rival Anthropic.
The announcement late Friday came just hours after President Donald Trump issued an executive order blacklisting Anthropic, whom he labeled a “national security risk.”
The rift between the Trump administration and Anthropic had been escalating for days and reached a boiling point Friday when Defense Secretary Pete Hegseth designated the startup a “supply-chain risk to national security.” The rare and severe label, typically reserved for foreign adversaries, effectively bars any federal agency or Department of Defense contractor from using Anthropic’s technology.
The conflagration follows weeks of stalled negotiations over the military’s use of Claude. Late Friday, President Trump took to Truth Social to demand all agencies “IMMEDIATELY CEASE” operations with the company, accusing it of obstructing national interests.
“We don’t need it, we don’t want it,” Trump posted, threatening “major civil and criminal consequences” if the company failed to comply with the phase-out.
At the heart of the dispute are red lines in the ethical application of AI. Anthropic had reportedly refused to grant the military unconditional access to its models, seeking written assurances that the technology would not be used for domestic mass surveillance of American citizens, or the development of fully autonomous weapons systems capable of using lethal force without human intervention.
While Anthropic maintained these boundaries were essential safety protocols, the Pentagon argued that private suppliers cannot dictate the terms of lawful military operations.
OpenAI CEO Sam Altman successfully navigated the political minefield by announcing a deal with what he pointedly referred to as the “Department of War.”
Altman claimed OpenAI secured the same safety red lines Anthropic sought, including prohibitions on surveillance and a requirement for human responsibility in the use of force, but framed the agreement as a partnership rather than an ultimatum.
“In all of our interactions, the DoW displayed a deep respect for safety,” Altman posted on X. He noted that OpenAI would implement “technical safeguards” and deploy personnel to ensure the models behave as intended.
Industry analysts suggest OpenAI’s success may stem from a “savvier” approach to Washington. While Anthropic lobbied against federal provisions in the Big Beautiful Bill and clashed with officials over regulation, OpenAI has worked closely with the government’s GenAI.mil initiative, arguing that its existing internal safeguards are robust enough to allow for “all lawful uses.”
Anthropic has vowed to challenge the “supply chain risk” designation in court, calling the move an act of “intimidation.”
The legal battle could have far-reaching implications for the AI industry; the designation threatens Anthropic’s enterprise revenue, as many private sector clients also hold contracts with the U.S. military.
As OpenAI moves to integrate its models into the nation’s most sensitive networks, the industry remains divided. While many in Silicon Valley applauded Anthropic for its moral standing, others viewed OpenAI’s agreement as the pragmatic path forward in an era where AI is increasingly viewed as the ultimate tool of national defense.
A labor coalition representing more than 700,000 tech workers is demanding that industry giants Amazon.com Inc., Google, Microsoft Corp., and OpenAI follow Anthropic’s lead in resisting recent Pentagon mandates. According to a Bloomberg report, the petition urges the companies to reject military contracts like the Pentagon’s that require stripping safety protocols from frontier AI models.
Pressure from the rank and file comes as OpenAI faces a growing public backlash and a “Cancel ChatGPT” movement. Critics and former users are fleeing the platform following Altman’s decision to accept a Department of War contract. The coalition’s letter explicitly warns against the ethical risks of these partnerships, urging their respective employers to refuse compliance should they or the labs they fund enter into future agreements involving lethal or surveillance-heavy military operations.
The fallout has already begun to impact OpenAI’s bottom line, with more than 700,000 users reportedly abandoning ChatGPT in protest. As the divide between Silicon Valley labor and defense objectives widens, the industry faces a critical turning point regarding the militarization of AI.
