What I said is pertinent: Anthropic refusing to build software for autonomous weapons platforms is a national security issue. They have the capability, but they just don't want to do it because they don't feel like it.
Soldiers don't need to die because a company is stubborn.
I think the whole problem is that they don't. Their systems are wholly unsuitable for _fully_ automated killing. Random schools and fishermen don't need to die because an administration stubbornly refuses to properly confront reality.
Anthropic isn’t just used for image analysis, and is actually a bad option as you said. It’s used mostly for programming, and what happens when it finds “missile system” in your prompt and refuses to continue coding?
I can answer your direct question if you'd like, but I'm actually missing what I think is some larger point in why you asked it. Would you be okay elaborating please?
Your argument would be more reasonable if you didn't handwave genuine and difficult moral and ethical decisions under those dismissive terms.
Civilians, including certainly 120 school girls, don't need to die because an AI hallucinated. Anthropic engages its responsability if their tools are used to designate targets. They have a right not to be implicated - or the Trump administration has to be open about requisitioning their tools and drafting their engineers in the name of a warm against Iran, which is another status altogether.
As it stands, there is no war against Iran, especially not one approved by Congress, and the only ones putting soldiers at risk are a senile reality show conman who was elected on the explicit promise of no new wars, and his gung-ho, racist, alcoholic TV host secretary of defense.
> Your argument would be more reasonable if you didn't handwave genuine and difficult moral and ethical decisions under those dismissive terms.
Anthropic willingly got itself into bed with the Pentagon. What did expect to do there, use their models to find the best locations to plant flowers?
As a company, you are to do what the client asks you to do or have your contract cancelled. Now it is bad luck that Anthropic decided not to do what the US government asked it to do and it is suffering the consequences. Maybe they should have used those fancy models to figure out that if they decide to say no to something, then every US government agency can say no to them. Suing to do business with the government again is poor business.
A judge reviewed the case for weeks, has training and experience in the law, had access to all relevant documents from both parties; the latter got the opportunity to argue their point at length, and the judge who knows the case better than you decided both the government and you are wrong.
Maybe things are a little more complicated than your simple rules.
Soldiers don't need to die because a company is stubborn.