According to Axios, Hegseth issued an "ultimatum" this week during a meeting with Dario Amodei, CEO of Anthropic, at the Pentagon, demanding that the company agree by Friday to allow the U.S. military unrestricted use of its model Claude, or else the Department of Defense would consider cutting ties and labeling it a "supply chain risk," even pushing to invoke the Defense Production Act to force it to adjust the model according to military needs. U.S. officials said the military wants to free itself from corporate restrictions on specific uses while also worrying about losing Claude, a key model currently used for the most sensitive and classified tasks. The report said the controversy centers on Anthropic's two red lines: opposition to using it for mass surveillance of Americans, and opposition to using it for fully autonomous lethal weapons. Hegseth emphasized that he would not accept companies placing limits on the "lawful use" by the military. The two sides also exchanged arguments over rumors that Claude was used in an operation through a collaboration with Palantir, which Anthropic denied having raised any objections to the partner. The meeting included several high-ranking officials from the Department of Defense and the general counsel, indicating that the Pentagon is preparing to escalate measures while also assessing the feasibility of alternatives such as OpenAI, Google, and xAI.
Image source: network
Original article: toutiao.com/article/1858077995440329/
Statement: This article represents the views of the author.