This article first appeared on GuruFocus.
Anthropic is suddenly at the center of a high stakes standoff with the Pentagon. U.S. Defense Secretary Pete Hegseth has reportedly given the company until Friday to loosen safeguards on its Claude AI model, warning that the government could escalate the matter under the Defense Production Act if it refuses.
According to Axios, Claude is currently the only AI model being used for classified Pentagon operations. But Anthropic has drawn clear red lines, including restrictions against domestic mass surveillance and autonomous weapons systems. That tension came to a head Tuesday when Pentagon leaders met with CEO Dario Amodei. Officials reportedly warned that Claude could be removed from defense workflows, and contractors pressured to follow suit, if the safeguards remain in place.
The Pentagon is also exploring alternatives from OpenAI, xAI and Alphabet’s (NASDAQ:GOOG) Gemini, though Claude is said to still lead on performance. One defense official bluntly told Axios, The only reason we’re still talking to these people is we need them.
