Pentagon bans Anthropic, taps OpenAI
In a major policy shift, the Trump administration has ordered federal agencies to stop using Anthropic's AI, citing supply chain risks. Just hours later, the Pentagon struck a deal with OpenAI to provide its AI systems, cementing OpenAI's position as the U.S. government's preferred AI partner.
The dispute escalated after the Pentagon demanded that Anthropic loosen its ethical guidelines, which prohibit the use of its AI for mass surveillance and fully autonomous weapons. Anthropic refused, stating it could not "in good conscience accede to their request." In response, President Trump accused the company of trying to "STRONG-ARM the Department of War." Defense Secretary Pete Hegseth subsequently designated Anthropic a "supply-chain risk to national security," a label typically reserved for foreign adversaries. This designation immediately bars any contractor, supplier, or partner doing business with the U.S. military from also conducting any commercial activity with Anthropic. The company has stated it will challenge the "unprecedented action" in court. The Pentagon had a $200 million, two-year agreement with Anthropic, whose Claude AI was the only model deployed on the Pentagon's classified networks. Federal agencies have been given a six-month transition period to phase out Anthropic's technology. Hours after the public falling out, OpenAI's CEO Sam Altman announced a new deal to deploy its models on the Pentagon's classified networks. Curiously, Altman stated that the agreement with the Pentagon includes the same "red lines" that were at the heart of the dispute with Anthropic: prohibitions on domestic mass surveillance and a requirement for human responsibility in the use of force. Unlike Anthropic, OpenAI does not currently have the infrastructure to operate on classified military networks. The new agreement will require a significant technical integration process, including security certifications and personnel clearances, before its models can be deployed. The move has drawn support for Anthropic from within Silicon Valley, including from competitors. Nearly 500 employees from OpenAI and Google signed an open letter stating "we will not be divided," and noted that the Pentagon is also negotiating with Google to agree to the terms Anthropic refused.
