India Questions Home-Help Apps Over Data for Physical AI Training

India’s IT ministry has sent inquiries to home-help service platforms, asking how video and sensor data from gig workers is collected and how consent is obtained when that operational footage is used for physical-AI training. The review examines the shift from service-delivery recordings to machine-learning datasets, raising questions about labor protections, privacy safeguards and purpose-limitation compliance.

The ministry’s focus is on the boundary between data gathered for task execution and data repurposed to train models that replicate physical tasks. Platforms that deploy workers to homes and businesses routinely capture visual and sensor streams. When those streams feed AI development, the original consent framework may no longer apply, creating gaps in transparency and worker awareness.

The scrutiny brings together concerns relevant to human resources, legal compliance and healthcare-adjacent services, as well as government oversight of emerging training-data supply chains. The inquiry signals that regulators are examining how consent, data-minimization principles and purpose-limitation rules hold up when operational footage becomes a training asset for physical AI systems.