[ExI] AI for the Pentagon (was: Re: ai in education)
Ben Zaiboc
benzaiboc at proton.me
Sun Mar 8 09:30:33 UTC 2026
> On Sat, Mar 7, 2026, 8:34 PM spike jones via extropy-chat <extropy-chat at lists.extropy.org> wrote:
>
> The military needs something that doesn't need to be intelligent so much as it needs to be able to coordinate and control … the golden dome.))
>
>
> >…Then why is it asking Anthropic?
>
>
> My best guess is the US military wanted Anthropic to develop a system to do what it needed.
>
>
> >… That's not Anthropic's specialty…
>
>
> Looks like that was also the military’s conclusion.
We seem to have wandered from the original question, and premises.
The US military want an AI system to perform automated, detailed mass surveillance of american citizens and autonomous targeting of and firing on designated targets, with safety guardrails removed to allow these things, was my understanding.
Consider what this enables, in terms of combatting domestic 'terrorism' (defined in any way the current administration wants. And remember that if one country gets a certain technology, it's not long before other countries get it too, and also remember that places like Saudi Arabia, for example, define atheists as terrorists).
That's a far cry from a missile defence controller, which is a much simpler (although still complex) requirement.
What the pentagon are asking for is terrifying, and I think the people at Anthropic realised that.
--
Ben
More information about the extropy-chat
mailing list