In a sprawling textile plant in Karur, a city in the southern Indian state of Tamil Nadu, workers are being handed GoPro cameras and smart glasses as part of a quiet experiment in automation. The devices capture their every motion on the factory floor, according to reporting by NPR. The footage is not being collected for quality control or training videos for new hires. Instead, it is being used to teach artificial intelligence systems and robots how to perform the same tasks, raising uncomfortable questions about the future of human labor in global supply chains.
The practice is unfolding at Texworth, a textile manufacturer that produces goods for international brands. On a recent visit, three men walked the length of the factory handing out recording equipment to employees. Among those asked to wear a GoPro was Shoba Kirubakaran, a 40-year-old worker who was busy packing yellow aprons decorated with chicken motifs into plastic wrapping. The aprons were destined for Designs By Danica, a brand that sells home and lifestyle products. For Kirubakaran, it was not the first time she had been told to strap on a camera. She said it was the third occasion she had been asked to do so.
What she did not know, until NPR told her during an interview, was the purpose of the recordings. The footage was being gathered to train AI-powered robots to eventually take over her job. "Nobody told me anything about it," she said. Her reaction captures a growing tension in the global economy: the same workers whose skills are being digitized are often the last to know that their livelihoods are being used as training data for their own replacements.
The arrangement raises immediate ethical and legal questions. In many countries, employers are required to inform workers when they are being recorded, particularly if the footage could be used for purposes beyond routine monitoring. India's labor laws offer some protections around privacy and consent, but enforcement in factory settings is uneven. Workers like Kirubakaran, who may have limited access to legal resources or fear retaliation, are especially vulnerable. The fact that she had already worn the camera multiple times without understanding its purpose suggests that whatever consent process exists at Texworth is not functioning as intended.
The broader context is a global race to automate manufacturing. Companies in the United States, Europe, and Asia are investing heavily in robotics and AI as a way to reduce labor costs, improve consistency, and insulate production from human error or absenteeism. Textile work, which involves repetitive hand movements such as folding, packaging, and stitching, is an attractive target for automation. Robots that can replicate those motions could operate around the clock without breaks, benefits, or wages. The data collected from workers like Kirubakaran is the raw material that makes such systems possible.
Yet the technology is not there yet. Training a robot to match the dexterity and adaptability of a human hand remains extraordinarily difficult. Every fold, every twist of the wrist, every adjustment for a slightly different fabric requires thousands of examples. That is why companies are turning to real workers on real production lines. By recording employees performing their tasks, engineers can build datasets that teach machines how to mimic human motion. In effect, the workers are training their own successors, often without realizing it.
For the workers themselves, the implications are stark. If the automation succeeds, the jobs that sustain entire communities in places like Karur could disappear. Textile manufacturing is a major employer in Tamil Nadu, particularly for women. Many workers have spent decades on factory floors, developing skills that are now being captured on camera and converted into code. They may not have the education or resources to transition into other industries. The promise of new jobs created by automation often rings hollow for those whose existing jobs are eliminated.
The company's perspective is not included in the NPR report, leaving open questions about Texworth's intentions and whether it believes the automation will lead to layoffs or simply to new roles for human workers. Some manufacturers argue that robots will take over only the most repetitive and physically demanding tasks, freeing humans for more complex work. But in practice, the line between repetitive and skilled work is blurry, and the pace of displacement often outruns the creation of new opportunities.
There is also the matter of transparency. Workers who are recorded without full knowledge of how their data will be used cannot meaningfully consent. Even if a company discloses that footage will be used to "improve processes," that language may obscure the ultimate goal of replacing human labor. Ethicists and labor advocates argue that workers should have a say in how their movements and likenesses are used, particularly when the end result could be their own unemployment.
The Karur case is not isolated. As AI systems become more capable, similar experiments are likely to proliferate across industries and geographies. Warehouses, farms, meatpacking plants, and assembly lines are all candidates for this kind of data collection. The workers involved are often low-wage, non-unionized, and unaware of the long-term consequences. Without stronger oversight, the transition to automation could deepen existing inequalities rather than distribute its benefits broadly.
For Kirubakaran, the immediate concern is simpler. She has a job to do, and she has been asked to wear a camera while doing it. She did not know that her movements were being used to build a machine that could one day perform them without her. When NPR informed her, she expressed surprise and unease. Her story is a reminder that the future of work is not being built in Silicon Valley laboratories alone. It is being assembled, frame by frame, on factory floors in places like Karur, by workers who may never see the robots they are helping to create.
(0)