Anthropic says its latest AI model is too powerful for public release and that it broke containment during testing

Anthropic said Claude Mythos is too good at hacking and that's why you won't be able to use it anytime soon.