Industry Seeks Narrower AI Labelling Rules, Warns Against Overreach

India’s technology industry has urged the government to narrow the scope of its proposed amendments to the IT Rules, 2021, which mandate visible labelling and watermarking of AI-generated and synthetic content. In detailed submissions to the Ministry of Electronics and IT (MeitY) on November 8, leading associations, Nasscom and Business Software Alliance, said that while the intent to curb deepfakes and deceptive content is welcome, the current draft risks covering nearly all AI-assisted material online.

Nasscom cautioned that the broad definition of "synthetically generated information" could include benign, everyday uses of AI in text, design, or video editing. Both bodies recommended limiting the labelling requirement to "deceptive or harmful" synthetic content that poses real risks to users or public trust.

They also sought flexibility in compliance, suggesting that metadata-based traceability and provenance tools should be prioritised over rigid, 10% visible labels or watermarks, which could disrupt user experience and increase compliance costs. Instead of prescriptive markings, Nasscom proposed a harm-based, outcome-driven framework consistent with global standards such as the Coalition for Content Provenance and Authenticity (C2PA).

Both groups have also called for exemptions for enterprise, backend, or non-consumer AI applications, including analytics, restoration, or editing tools, and for text-only content, where visible labelling may be impractical. They further urged the government to test implementation challenges through voluntary sandboxes before any full-scale rollout.

While expressing support for MeitY’s broader goal of AI transparency, industry representatives cautioned that excessive labelling could stifle innovation and overburden intermediaries. "The focus should be on intent and harm, not every instance of AI use," one submission noted.

Despite industry feedback, NDTV Profit earlier reported that the government may not fully roll back its proposed rules but is likely to ease the 10% disclosure threshold to around 5–7%, according to sources. CII and Jetsynthesys' Rajan Navani had earlier issues similar comments, complimenting the intent but questioning the practicality of broad-based AI regulations.