Anthropic - we have a model that’s so dangerous that we had to turn it down! OpenAI - we have a model that’s so dangerous that it found a way to escape itself!!! I think I already saw this elsewhere…
