The training of artificial intelligence systems has become increasingly dependent on human labor, but a troubling pattern is emerging: much of this work is being outsourced to gig workers in the Global South at minimal pay. Zeus, a medical student in Nigeria, exemplifies this trend—he records videos of himself performing physical movements from his apartment, using just an iPhone and ring light, to train humanoid robots. This work represents a critical yet largely invisible component of AI development, where companies like Figure AI and others source motion-capture data from remote workers who are compensated far below what similar work would cost in developed nations.
The policy implications are significant. Unlike traditional manufacturing or tech outsourcing, this labor exists in a regulatory gray area where worker protections, minimum wage requirements, and labor standards remain unclear. Tech companies argue this approach democratizes opportunity for workers in regions with limited job markets, while critics contend it perpetuates exploitative practices by leveraging global wage disparities. The lack of transparent disclosure about these labor practices means consumers and investors often remain unaware of the true supply chain behind the AI systems they use.
As the humanoid robotics industry accelerates toward commercialization, policymakers face mounting pressure to establish clear labor standards for AI training work. The situation mirrors earlier debates around content moderation and data annotation, where similar outsourcing practices eventually prompted calls for regulation. Without proactive policy intervention, the AI boom risks systematizing a two-tiered labor market where essential training work remains systematically undervalued based on geography.