Brazilian workers, housework video AI training... payment only for approved portions

Brazilian workers are increasingly earning money by filming household chores for AI and robotics training, with payment based on the length of approved footage rather than hours worked, according to G1. Participants mount phones on their heads or chests to capture first-person footage of washing dishes, cleaning and cooking, submitting clips through platforms such as the Minute Data app after recruitment by separate services. The model mirrors robotics experiments that have trained machines on up to 1 million hours of human video. The arrangement matters because it turns everyday labor into a paid data pipeline for embodied AI, while shifting compensation risk onto workers whose clips must pass platform review. What to watch next is whether approval rates and per-minute pay hold up as demand for first-person training data grows.