A garment factory floor in Gurugram. Rows of workers sit in front of sewing machines, stitching fabric, folding garments and aligning seams with practiced precision. At first glance, it looks like any ordinary factory in India.
But there is something unusual happening here. Several workers are wearing small cameras strapped to their heads, pointed directly at their hands.
The footage from those cameras is not being recorded simply for productivity checks or workplace monitoring. It is being used to train Artificial Intelligence systems and humanoid robots.
That is the deeper story quietly unfolding inside parts of India’s manufacturing sector.
According to a detailed investigation by Scroll.in, workers at a Gurugram unit of apparel manufacturer Pearl Global Industries were asked to wear camera devices while working. When reporters spoke to workers, many said they had been told the cameras were meant to observe their work process and efficiency during shifts.
None of the workers interviewed appeared to fully understand that their hand movements and actions could later become training material for AI-powered robotics systems.
The incident has sparked larger questions about labour rights, informed consent and the future of automation in India.
Why AI companies want factory workers
The modern AI race is no longer only about chatbots or software. Technology companies around the world are now trying to build humanoid robots capable of performing human tasks such as folding clothes, sorting products, packing boxes or assembling components.
For robots to learn these actions, they need enormous amounts of human behavioural data.
This is known as “egocentric” or first-person data. Instead of recording a person from across the room, the camera captures exactly what the worker sees while performing a task. AI systems then analyse tiny details such as wrist movement, finger grip, hand speed and coordination.
Using a machine-learning process called imitation learning, robots attempt to replicate these movements.
Industry researchers increasingly believe that collecting high-quality real-world human motion data has become one of the biggest challenges in robotics development.
That is why factories and workers are becoming valuable sources of AI training material.
The startup behind the data collection
The Scroll.in investigation linked the Gurugram operation to Egolab.AI, a startup reportedly founded in early 2026 by two young entrepreneurs from Maharashtra.
The company described itself as a first-person data collection platform focused on industrial and robotics training datasets. According to company materials referenced in reports, lightweight wearable cameras were used to collect footage from workers performing factory tasks.
The startup was later acquired by Build Artificial Intelligence Inc., a US-based company registered in Delaware.
Build AI subsequently released a massive open-source industrial robotics dataset on Hugging Face, a major AI platform used globally by developers and researchers. The dataset reportedly included around 100,000 hours of first-person industrial footage and more than one billion image frames collected from real-world work environments.
The companies named in reports have not publicly stated that workers were intentionally misled. However, the debate centres on whether workers truly understood how their data would be used.
The real concern is consent
This issue is not simply about cameras in factories.
The bigger question is whether workers were given proper informed consent.
There is a major difference between telling workers that cameras are monitoring workplace activity and explaining that their movements may be used to train robots capable of performing similar jobs in the future.
Labour experts argue that many workers may not fully understand the long-term implications of such data collection, especially in sectors where wages remain low and job security is limited.
India currently does not have a dedicated legal framework specifically governing biometric movement data collected for AI training purposes in workplaces.
The Digital Personal Data Protection Act, 2023 provides a broader structure for personal data governance, but experts say regulations around wearable workplace AI data remain unclear and underdeveloped.
This has created a grey area where technology companies can move faster than labour regulation.
India’s workers powering global AI
The global AI data economy is growing rapidly.
Research firms estimate that the humanoid robotics industry could expand dramatically over the next decade as companies invest heavily in automation for factories, warehouses and logistics operations.
To build these systems, firms require vast amounts of real-world human behavioural data.
India’s large workforce and comparatively lower labour costs make it attractive for such data collection operations.
Similar work has already emerged in sectors like content moderation, image labelling and AI data annotation, where workers in developing countries help train systems used by global technology companies.
Now the same model appears to be expanding into physical labour and human movement.
Workers earning modest monthly salaries may unknowingly contribute to datasets that later help create machines worth thousands or even hundreds of thousands of dollars.
Critics argue that this creates a major imbalance in value distribution.
The workers provide the skills and behavioural knowledge. The long-term commercial benefits flow largely to technology companies abroad.
The argument from technology companies
Companies working in robotics and AI generally argue that automation is inevitable and that collecting such data is necessary for technological progress.
Some firms also claim that AI and robotics will create new jobs in robot supervision, maintenance and operations.
Supporters of the industry say India could benefit economically by becoming an important hub for AI training infrastructure and robotics development.
They also note that not all companies operate identically. Some reportedly disclose the purpose of data collection more clearly and provide better worker agreements.
However, critics say the core issue remains transparency.
Workers deserve to know exactly how their labour and behavioural data are being used.
A debate India has barely started
The controversy surrounding wearable cameras in factories has opened a much larger conversation about the future of labour in the AI age.
Should workers receive royalties or compensation if their data helps train profitable AI systems?
Should employers be legally required to explain how AI training data will be used?
Should India create specific rules governing workplace biometric and behavioural data collection?
These questions are now becoming increasingly important as AI moves beyond software and into physical workspaces.
Because the real story is not only about robots.
It is about human knowledge being quietly converted into digital assets, often without workers fully understanding the process.
The machines learning these tasks may eventually power factories around the world.
But the human hands teaching them belong, increasingly, to workers sitting in Indian factories today.
Subscribe Deshwale on YouTube
