MeitY Reviews House Help Platforms Data Collection Process For Physical AI Training

The Ministry of Electronics and Information Technology (MeitY) has sought details from home services platforms, including Pronto, Snabbit and Urban Company’s InstaHelp, over the collection of first-person video and sensor data from gig workers for training physical AI models.

As per a report by ET , a meeting chaired by MeitY secretary S Krishnan yesterday brought together senior executives from the quick home services platforms and other technology and robotics companies to discuss concerns around workers’ consent, consumer privacy and data sovereignty.

Officials are concerned that recordings of everyday household activities could infringe workers’ rights, compromise consumer privacy and pose risks to national data security.

The ministry sought details on the standard operating procedures (SOPs) followed by companies for collecting, processing and storing such data. It also sought information on safeguards for obtaining workers’ consent and protecting the privacy of other individuals captured in the recordings.

The ministry also sought clarity on whether companies were selling, sharing or transferring the collected data to physical AI developers outside India. It asked Pronto to submit details of its data collection practices and the SOPs it follows, ET reported, citing people familiar with the discussions.

Inc42 has reached out to Snabbit, Pronto and Urban Company for their comments on the development. The story will be updated upon receiving responses.

Concerns were also reportedly raised about India becoming a source of cheap training data for foreign AI companies. Executives suggested that the country should focus on developing its own physical AI capabilities rather than merely supplying data to overseas developers.

This comes shortly after Pronto faced scrutiny over a pilot programme to record household chores using cameras worn by service professionals to generate training data for robotics systems.

The controversy also brought Snabbit and Urban Company under scrutiny over their potential involvement in this. Snabbit subsequently clarified that it had conducted a limited assessment of the tech in a controlled training-centre environment but decided against proceeding with the proposal. 

Urban Company founder Abhiraj Singh Bhal said the company neither records inside customers’ homes nor plans to do so.

Pronto, meanwhile, confirmed that it was running an opt-in pilot involving first-person video recordings of household chores. Initially, it said the feature was available only to customers who actively opted in during booking and was limited to less than 0.01% of its users.

The startup later told Inc42 that raw footage was deleted immediately after the work was completed, while an anonymised, end-to-end encrypted recording remained accessible only to customers for 48 hours. 

The anonymised footage was subsequently used for AI training, as outlined in its terms.

This further casts a shadow on Pronto founder Anjali Sardana’s expanding ambitions in physical AI beyond the home services business.

In July, the startup formalised its data collection experiment through Verified, an opt-in premium service in which professionals wear head-mounted or body-mounted cameras during visits.

Last month, Sardana launched Provision Labs, a San Francisco-based physical AI venture that will collect data to train and test AI systems designed to perform real-world tasks using robots and other machines.

Provision Labs also announced the acquisition of Mumbai-based Basilisk Industries, a startup focused on perception technologies. Basilisk founder Shourya Watwani will lead Provision’s hardware efforts and work on its in-house post-processing models.

The relationship between Pronto’s household data collection and Provision Labs’ operations remains unclear. 

The post MeitY Reviews House Help Platforms Data Collection Process For Physical AI Training appeared first on Inc42 Media.