Protecting producers interests | Photo Credit: David Gyung

The AI economy is evolving from chat bots and conversational agents to instructing Robots to do physical tasks — aka physical AI — encompassing labelling, annotation, Reinforcement Learning from Human Feedback (RLHF), and prompt engineering.

Firms, mostly start-ups including HumynAI Labs, Egodata, and Neo Cambrian have deployed gig workers on the ground to label for physical AI and robotics AI to perform tasks such as housekeeping, plumbing, painting, and decorating to name a few, so that data from these processes are captured for training the models.

To enable this, digital platforms such as Encord, Labellrr, and Micro1 hire thousands of gig workers around the globe, equip them with depth sensing cameras and other accessories including Light Detection and Ranging (LiDAR) sensors to scan and annotate the physical world to be used by the physical AI and robotic companies. AI doesn’t consist of only algorithms, but encompasses the invisible human labour beneath it.

Workers’ weak power

These millions of gig workers who produce and feed content in to the AI algorithms often operate in an oligopsony market, where there are few physical AI firms procure their annotated content. The workers are often in a structurally weak position, and hence have a weak bargaining power with respect to wages, security and safety, amongst others.

While the regulations worldwide are focussed on monopoly or oligopoly markets, focusing on consumers and their welfare, privacy and data protection, they often are inadequate to address producer concerns. It is in this context that our research (accessible at: https://papers.ssrn.com/sol3/Delivery.

cfm?abstractid=6716218) provides insights in to the following important, often overlooked regulatory aspects.

First, is the importance of review and rating as signals of quality of work and workers in the above marketplaces. Platforms normally have a reputation system through which customer feedback about the quality of work carried out by the gig worker is routed, curated and published. Bayesian rating applied by most of the platforms require a threshold number of reviews to be materially effective.

Our research shows that a minimum of 800 transactions are required for the reviews to have any significant impact. This minimum threshold may weed out entrants and benefit only the first movers.

Further, most of the gig workers multi-home or switch platforms and if their reviews and ratings from one platform are not carried over to the other, they are disadvantaged. Hence regulators mandate portability of worker reviews across platforms, much the same way as the consumer data portability as mandated under the European Union Digital Markets Act.

Commission factor

Second is commission rates charged by the platforms, normally on the less price sensitive side — producer firms in this case. While the producer firms lobby for a commission rate cap to protect their profits, it is imperative that the regulators analyse the effect of such caps on the platform’s ability to invest. Caps set too low may put pressure on the investment budgets of the platforms on infrastructure, cyber security and fraud prevention resulting in corresponding reduction in quality of artefacts in the marketplace.

Globally, no regulators have yet prescribed a cap on commission rates on digital platforms, we caution against such caps on the producers especially in oligopsony markets where the negative effects may be significant.

Third, is the regulatory frameworks designed to protect consumers through quality assurance standards. Dictating minimum quality standards especially on the artefacts produced by the gig workers, increase compliance costs, credentialing burdens, and approval timelines, thereby discouraging newer gig workers to be part of this gig economy. Regulatory interventions on such minimum quality levels, though very much required in an oligopoly market, will have negative consequences on the workers in an oligopsony market and hence require calibrated interventions.

Fragmented regulation

Regulation of the digital and AI economy is very fragmented. While content and cyber security of the digital platforms are handled by the Ministry of Information Technology through the IT Act and Digital Data Protection Regulation Act; welfare protection (of the consumers) is legislated through the Ministry of Consumer Affairs, Food and Public Distribution through the Consumer Protection (E-Commerce) Rules.

Competition Commission of India, on the other hand regulates market competition and contestability issues through the Digital Competition Bill. It is time that a unified cohesive and comprehensive framework is evolved for resolution of regulatory issues of digital platforms, that spans across various areas of labour economy.

Overall, India should proactively build a producer-centric AI platform regulatory framework before harms become entrenched.

Sridhar and Rao are Professors at IIIT-Bangalore

Published on June 12, 2026