India urged to craft bespoke AI safety rules instead of copying US pact

India must craft its own AI safety framework tailored to its languages, scale and application‑led ecosystem, industry experts said, rather than copy the voluntary pact signed by US President Donald Trump and leading tech firms. The White House Accord on Super Intelligence, signed on September 29 by Trump and executives from Google, OpenAI, Anthropic, Meta, Nvidia and Elon Musk’s xAI, commits the companies to internal controls, dedicated oversight teams, independent external audits, board‑level oversight and regular meetings on safety standards. Nasscom Chairman and Fractal Group CEO Srikanth Velamakanni warned that the US accord is mainly a self‑policing push and urged India to use a domestic commitment as a stepping stone toward global AI safety understanding. He stressed that as models become more capable and autonomous, safety must stay ahead of capability to avoid catastrophic risks. Velamakanni contrasted the US approach with the EU’s binding risk‑based rules and China’s state‑directed controls, noting India’s role in the Global Partnership on AI and the AI Impact Summit in February as opportunities to prevent worst‑case outcomes without denying the Global South access to frontier models. Counterpoint Research analyst Soumen Mandal said India can draw inspiration from the US pact and speed up building a trusted AI ecosystem, covering pre‑deployment testing, independent audits, serious‑issue reporting and clear responsibility for AI risks. Grant Thornton Bharat partner Amit Khanna added that any framework must suit Indian market conditions, ensure reliability across vernacular languages and avoid exploiting knowledge asymmetry. Gnani.ai CEO Ganesh Gopalan called for a level playing field, insisting global frontier models follow the same rules as Indian ones. While Velamakanni argued voluntary self‑regulation is insufficient and called for binding rules on transparency, incident reporting and accountability, Mandal and Khanna favoured a hybrid model: mandatory rules for defence, healthcare, banking and government, with voluntary standards for fast‑moving technical areas. The Indian government is already considering AI regulation; IT Secretary S Krishnan and IT Minister Ashwini Vaishnaw have signalled that existing laws may need to be updated to address deepfakes and synthetic content.

Context

India’s push for a home‑grown AI safety regime reflects its large, multilingual user base and growing AI‑driven businesses, while balancing the need to stay competitive with global tech leaders.