Pentagon Summons Anthropic CEO Over A.I. Use on Classified Systems

Pentagon officials summoned Dario Amodei, Anthropic’s chief executive, to the Pentagon to meet Defense Secretary Pete Hegseth and discuss how the company’s technology is used on classified systems. The two sides had agreed to a $200 million pilot contract last year, but a Jan.

9 memo from Mr. Hegseth prompted renegotiations over restrictions on the model. The department has signed an agreement with Elon Musk’s xAI and is close to a deal with Google, steps officials hope will give them leverage with Anthropic. Pentagon negotiators say contracts must let the department use the models as it sees fit, within the law, while allowing companies to build safety provisions into their models, which the companies call the “safety stack.” Anthropic, the first company authorized to work on the military’s classified networks, says it is willing to loosen restrictions but has demanded guardrails to prevent mass surveillance of Americans and the deployment of autonomous weapons with no humans in the loop.

United States

anthropic, dario amodei, pentagon, pete hegseth, classified systems, 200 million, safety stack, guardrails, mass surveillance, autonomous weapons