The chapter explores research on model training, focusing on overcoming the reversal curse with doubled data. It discusses a new model, Cobra, and its application for efficient inference in language models. Furthermore, it touches on the UN General Assembly's resolution on AI governance and the implications of AI surveillance on society.