Startups warn AI approval rule could slow LLM development

India’s government has told companies that AI models, LLMs, generative AI software and algorithms in testing, beta development or deemed unreliable must seek “explicit permission of the government” before deployment for users on the Indian internet. The March 2 directive has drawn sharp criticism from startups, venture capitalists and experts, who say approval requirements could favor large corporations with resources for testing and compliance.

KissanAI founder Pratik Desai said the rule could hurt builders of foundational and fine-tuned models, calling it “License Raj 2.0” and questioning who would evaluate models. Capstone Legal’s Ashish K Singh warned that restrictive regulation could push Indian IT companies to develop AI products overseas. Shorthills AI co-founder Paramdeep Singh said excessive red tape could disadvantage open-source developers in a fast-growing field.

Some industry voices backed the push for oversight. Krutrim said it would work with authorities to comply, while Vizzhy’s Vishu Vardhan supported regulatory policies for general AI and linked Hanooman to responsible AI. Others urged balance: Chaitanya Chokkareddy of Ozonetel said unclear definitions could create a slippery slope, and Harman’s Jai Ganesh said safeguards should not stifle research and entrepreneurship.