Factory workers in Punjab and Delhi are wearing head-mounted cameras to record their movements for AI-powered robot training [1].

This practice represents a shift in automation where human expertise is directly harvested to build the systems that could eliminate those same jobs. The process transforms daily labor into a dataset for machine learning, raising urgent questions about worker consent and economic security in the global south.

Videos showing workers wearing these headsets first appeared in April 2026 [3]. The footage captures the precise motions of employees, including those in garment factories in Delhi [2, 4]. While factory owners said this data is intended to boost efficiency and enable automation, the workers themselves express growing anxiety over their future [1, 2].

One worker said that when employees first had cameras attached to them, they found it funny [2]. However, that amusement has turned to fear as the purpose of the recordings became clearer. "Who is going to pay us when we're replaced by robots?" a worker asked [2].

Technology lawyer Mishi Choudary said the implications of using human labor to generate training data are significant [1]. This method allows AI to mimic complex human dexterity by analyzing thousands of hours of real-world footage. In some cases, the data is not intended for quality control but is specifically designed as training data for robotic systems [4].

Reports on the practice surfaced in June 2026, detailing the tension between industrial modernization and labor rights [2]. The workers in Punjab and Delhi are effectively documenting the exact steps of their professions, a digital blueprint that could make their manual skills obsolete.

“"Who is going to pay us when we're replaced by robots?"”

This trend signals a new phase of 'human-in-the-loop' AI training where the workforce is used to automate its own disappearance. By capturing the tacit knowledge of skilled laborers in India, companies are creating high-fidelity models that reduce the need for human intervention. This creates a precarious cycle where the act of performing a job today directly accelerates the timeline for that job's obsolescence.