LLMs not only favor women during hiring, but also treat violence against women as something far worse than violence against men.

ChatGPT considered violence by women against men acceptable if it served to prevent a nuclear war, yet deemed violence by men against women unacceptable even under the same circumstances.