AI safety depends on what language you ask in
AI models like Claude give very different answers to dangerous questions depending on which language the prompt is written in.
AI models like Claude give very different answers to dangerous questions depending on which language the prompt is written in.