MeitY Examines Home Services Platforms’ Data Collection for Physical AI Training

India’s Ministry of Electronics and Information Technology (MeitY) is examining how home services platforms collect video and sensor data from gig workers to train physical artificial intelligence (AI) systems. The ministry has sought information from companies including Pronto, Snabbit and Urban Company’s InstaHelp, focusing on worker consent, customer privacy, data security and the possible transfer of recordings to overseas AI developers.

The development follows growing interest in using first-person recordings of everyday household activities to train robots. These recordings can show how people clean floors, organise objects, fold clothes and perform other routine tasks. While such information could help machines learn to operate in real-world environments, collecting it inside private homes raises questions about who controls the footage, how it is stored and whether everyone captured in the recordings has agreed to its use.

MeitY Seeks Details From Pronto, Snabbit and Urban Company

MeitY convened a consultation with representatives from home services platforms and technology and robotics companies on November 12, 2026. The meeting was chaired by MeitY Secretary S. Krishnan.

According to reports by The Economic Times and Inc42, the ministry sought details about the standard operating procedures (SOPs) companies follow when collecting, processing and storing first-person video and sensor data.

Officials also wanted information about safeguards for obtaining consent from gig workers and protecting other individuals who might appear in recordings.

The ministry reportedly raised questions about whether collected data is being sold, shared or transferred to physical AI developers outside India.

Companies were asked to clarify their data-sharing arrangements, retention policies, vendor contracts and the nature of the information being collected.

Pronto was specifically asked to provide details of its data collection practices and related SOPs.

The consultation does not, by itself, establish that the companies have violated any law. It indicates that the government is examining an emerging data-collection practice as physical AI development expands in India.

Why Home Services Companies Are Collecting Video Data

Physical AI refers to artificial intelligence systems designed to understand and interact with the physical world. Unlike conventional AI applications that primarily process text, images or digital information, physical AI is intended to operate through machines such as robots and autonomous systems.

For a robot to perform household tasks, it needs to understand how objects move, how people interact with their surroundings and how physical actions are carried out.

Existing internet datasets cannot provide all the information required for these activities.

For example, a robot learning to fold clothes needs to understand hand movements, fabric behaviour, object positioning and the sequence of actions involved. Similarly, a robot designed to clean a kitchen needs information about different surfaces, utensils, appliances and unpredictable household environments.

First-person recordings can provide this information.

Workers wearing head-mounted or body-mounted cameras capture activities from their own perspective. Additional sensors can record movement and other physical interactions.

The footage can then be processed, labelled and used to train AI systems to recognise objects, understand actions and eventually reproduce tasks using robotic equipment.

This creates a potential commercial opportunity for companies that can collect large volumes of real-world human activity data.

Pronto’s In-Home Recording Pilot Attracts Attention

Pronto has emerged as the central company in the debate because it confirmed that it was conducting an opt-in pilot involving first-person video recordings of household chores.

The startup introduced a recording initiative through its Verified service, in which home service professionals wear head-mounted or body-mounted cameras while performing selected tasks.

The company said the pilot was limited to customers who actively opted in during the booking process. It also indicated that the initiative initially covered less than 0.01% of its users.

Pronto subsequently told Inc42 that raw footage was deleted immediately after the service was completed. An anonymised, end-to-end encrypted recording remained accessible to customers for 48 hours.

The anonymised footage was then used for AI training in accordance with the company’s stated terms.

The initiative has attracted attention because recordings inside homes can capture much more than the specific task being performed. They may also include household layouts, personal belongings, family members, visitors and other identifiable information.

Even when a worker and customer agree to recording, questions remain about how the privacy of other people inside the home is protected.

Snabbit and Urban Company Take Different Positions

The MeitY consultation also brought attention to how competing home services platforms are approaching physical AI data collection.

Snabbit clarified that it had evaluated the technology through a limited assessment in a controlled training-centre environment. The company said it decided against proceeding with the proposal.

Urban Company founder Abhiraj Singh Bhal has also said the company does not record inside customers’ homes and has no plans to introduce such a practice.

These positions distinguish the companies’ approaches from Pronto’s opt-in in-home recording pilot.

The issue is particularly important for home services platforms because customer trust is central to their business models. Consumers allow service professionals into private spaces, often while family members and personal belongings are present.

Introducing recording technology could create additional concerns unless platforms clearly explain what is being recorded, why it is necessary and how the resulting data will be used.

Worker Consent and Household Privacy Become Key Concerns

One of the main questions facing regulators is whether gig workers have meaningful control over recordings made during their jobs.

A worker may agree to wear a camera as part of a service assignment, but that does not automatically resolve questions about the privacy of customers and other people appearing in the footage.

There are several potential risks:

• Customer privacy: Recordings could capture personal belongings, household arrangements and private activities.
• Worker consent: Workers need clear information about recording, its purpose and how the footage will be used.
• Data retention: Companies must explain how long recordings are stored and when they are deleted.
• Third-party access: Data shared with external AI developers or technology vendors could create additional security risks.
• International transfers: Sending recordings outside India raises questions about data protection and accountability across jurisdictions.

Anonymisation can reduce privacy risks, but it does not automatically guarantee that footage cannot be linked to an identifiable individual. Faces, voices, locations and contextual details may reveal personal information if they are not adequately protected.

India’s Data Protection Framework Comes Into Focus

The scrutiny also brings India’s Digital Personal Data Protection Act, 2023, into focus.

The law establishes a framework for processing digital personal data and provides obligations relating to consent, transparency, data security and individuals’ rights.

Video recordings, facial images and location information may qualify as personal data when they identify, or can reasonably be linked to, an individual.

Companies collecting such information therefore need to consider whether their notices clearly explain the purpose of collection, how the data will be processed and whether it will be shared with other organisations.

The legal position can become more complicated when recordings include people who are not directly involved in the service booking.

MeitY’s consultation could help clarify how existing data protection requirements apply to emerging physical AI data collection practices.

However, the meeting should not be interpreted as an announcement of a new blanket prohibition on recording or using household footage for AI training.

India Risks Becoming a Supplier of Cheap Physical AI Training Data

Another concern raised during the consultation involved India’s position in the global physical AI industry.

Some executives reportedly questioned whether India could become a low-cost supplier of human activity data for foreign AI developers without developing its own advanced robotics capabilities.

Collecting real-world data is an important part of physical AI development, but it can also become a commercial industry in its own right.

Companies can collect recordings, remove identifying information, annotate movements and actions, and supply processed datasets to robotics developers.

The global demand for this information could create opportunities for Indian startups, particularly because the country has a large workforce and diverse real-world environments.

However, executives argued that India should also focus on developing domestic physical AI companies, robotics technologies and foundational capabilities instead of primarily exporting training data.

The distinction is significant. Building domestic technology businesses could allow India to capture more value from the emerging robotics industry rather than remaining limited to data collection and processing services.

Physical AI Creates a New Technology and Privacy Challenge

The growing interest in physical AI is creating a new category of data governance challenges.

Traditional AI training often relies on publicly available internet information, licensed datasets and user-generated digital content. Physical AI requires more direct observation of people interacting with real environments.

This makes data collection more difficult because recordings need to capture different tasks, movements, objects and conditions.

For example, a robot trained to fold clothes in a controlled laboratory may struggle with different fabrics, lighting conditions or irregularly shaped garments in an actual home.

Companies therefore need diverse datasets that reflect the unpredictability of everyday life.

However, collecting such information in private homes creates risks that are different from those associated with conventional online data collection.

MeitY’s latest consultation reflects the government’s attempt to understand these risks before the practice becomes more widespread.

The Bigger Picture

MeitY’s examination highlights the tension between India’s ambition to participate in the physical AI revolution and the need to protect workers and consumers. First-person recordings can provide valuable information for training robots, but household environments contain sensitive personal information that cannot be treated like ordinary industrial training data. Clear consent, limited collection, secure storage and transparent data-sharing arrangements will be essential if this business model is to expand responsibly.

The issue also has economic implications. India could become an important source of real-world training data for global robotics developers, but the government and industry must consider how to build domestic capabilities alongside data collection services. Developing Indian robotics companies, AI models and hardware could help the country capture greater value from the emerging industry while maintaining stronger control over its data ecosystem.

Looking Ahead

The next stage will depend on the information companies provide to MeitY and whether the government decides to introduce clearer guidelines for collecting and processing physical AI training data. Pronto’s recording pilot, along with the approaches taken by Snabbit and Urban Company, will remain important examples as policymakers assess worker consent, customer privacy and cross-border data transfers. Any future framework will need to balance innovation with meaningful safeguards for people whose activities are being recorded.

For India’s technology industry, the opportunity extends beyond home services platforms. Physical AI could create demand for robotics research, sensor technology, data annotation, machine learning and specialised manufacturing. However, sustainable growth will require companies to establish trust with workers and consumers. The ability to collect useful data responsibly could become a competitive advantage as India develops its position in the global robotics industry.