China’s cyber-space regulator has unveiled draft measures that would make companies responsible for the data used to train generative artificial intelligence (AI) models, such as Midjourney and ChatGPT.