Is that true?
So, Anthropic does not want to allow mass surveillance and autonomous killer machines.
The whole point is that this is a prerequisite for being able to get an AI contract with the U.S. military who is not into mass surveillance, but defence and war fighting using military means. Also, machines cannot have a Rank and therefore may not do life and death decisions. Actually, they may not do any decisions without approval of a dedicated soldier in charge.
That means, that an AI system may automatically drive a vehicle back based on military personal orders, that an AI can refill supplies by sending requests in-between logistics hubs, if given order by a surveying soldier in charge of the task, but it may not do independent decisions by the lack of a rank within the military.
To the opposite, do I wonder if Anthropic can guarantee which levels of security, safety and secrecy for the decision making and answering process from terminals and access points from within the U.S. military to it's Servers.