It seems like magic, but behind the technological revolution of AI lies an army of underpaid workers in low-income countries. To train large language models such as ChatGPT, thousands of ‘annotators’ must sift through hours of pornography and violent extremist content. Psychologist Angela Chukunzira condemns the harsh reality of an annotator. “They barely make a living”, she explains. Not only are their working conditions precarious, the psychological tolls are heavy; “When workers are permanently exposed to highly harmful content, they become numb.” OpenAI founder Sam Altman calls his technology “unimaginable at any previous time in history”, but the human exploitation behind it is dangerously familiar.