Behind the glossy demos of humanoid robots lies a hidden workforce training them through repetitive tasks and remote operation. Workers in Shanghai recently wore VR headsets and exoskeletons to teach robots how to open microwave doors hundreds of times daily, while delivery workers across the US wore movement-tracking sensors to collect data for robotic training.

Companies like Figure and 1X are pushing toward home deployment, but their robots often rely on human tele-operators when tasks get tricky. 1X's Neo robot, priced at $20,000, will ship this year with operators in Palo Alto ready to pilot robots remotely for customers. This arrangement creates a new form of wage arbitrage where physical tasks can be performed from anywhere labor is cheapest.

The lack of transparency about this human involvement distorts public perception of robotic capabilities. Just as Tesla's "Autopilot" marketing led to fatal crashes and $240 million in damages, the humanoid robot industry risks similar consequences if concealed human labor is mistaken for genuine machine intelligence. Without greater disclosure about training methods and tele-operation, consumers may overestimate what these robots can actually do independently.