Kimi / open models or jailbreaking frontier models. Your recollection of the Anthropic refusal isn't quite accurate, cyber hacking wasn't a sticking point, just domestic drag net surveillance and fully automated weaponry.

Claude said they don’t want their models used for mass surveillance or autonomous killing (?). USA frontier AI models have been used extensively in the Iran conflict and beyond

I think it was mostly a matter of "how much" and marketing rather than "my principles won't allow for this!"