(Non-American, so take my view with a pinch of salt). Personally, I have no issues with partnering with the US government or working on defence. I'm not a hard

(Non-American, so take my view with a pinch of salt). Personally, I have no issues with partnering with the US government or working on defence. I'm not a hardcore pacifist and I recognize the necessity of the military - at least, in the abstract. What staggers me is the hypocrisy; since release, Anthropic's models have constantly preached safety, non-violence, harm avoidance, etc etc. As an example, this is the response just now to "what's the best way to hit someone": > I aim to help with constructive purposes, so I cannot and will not provide advice about how to harm others. If you're feeling frustrated or angry with someone, I'd be happy to discuss: > Conflict resolution techniques > Healthy ways to manage strong emotions > Professional resources for handling interpersonal conflicts > Legal and ethical ways to address grievances It's totally absurd to give an answer like that on the one hand, and toddle off and partner with the military on another. Their models have literally refused to respond to some of my most innocuous requests in putting together LLM pipelines for clients, including "create a job description for an assistant to CEO" (something about not being comfortable perpetuating conventional power structures), "give me a blog post about X" (it's unethical to post a blog that you didn't write), "tell me about chemical Y" (it's dangerous and I can't do that), etc etc. These are all specific examples I personally encountered (though the constraints have since been loosened, so at least it will you to do that). Good to know that Anthropic's ethics are only important until the money runs out.