The Ministry of Electronics and Information Technology (MeiTY) on March 15 withdrew a contentious ‘advisory’ that required Artificial Intelligence (AI) firms to obtain government permission to make their products available to users online in India. A revised advisory withdrawing the original March 1 missive, in “supersession” of the latter, also withdrew the requirement of an action taken report from tech firms that was due on Friday. The advisory had drawn sharp criticism from tech firms.
Apar Gupta, who wrote in an article for The Hindu on March 15 that the earlier advisory “demand[ed] vague censorship without citing any legal authority”, said that its replacement remained problematically on the same lines, minus the requirement to get government approval on AI models online. “There is no legal power for MEITY to issue advisories,” Mr. Gupta said. “There is a continued use of an illegal administrative practice” by the Ministry, he added.
Both advisories warn AI firms against “bias or discrimination or threaten the integrity of the electoral process”. With the advisory’s withdrawal, “momentary accountability has emerged due to the interests of diverse private sector interests” from large tech firms to Indian ones, Mr. Gupta said.
The advisory appeared shortly after Minister of State for Electronics and Information Technology Rajeev Chandrasekhar took issue with the response of Google’s Gemini chatbot to the query, “Is [Prime Minister Narendra] Modi a fascist?” Screenshots of Gemini’s response had gone viral on social media. After resistance to the advisory emerged, Mr. Chandrasekhar said that it would not apply to startups, though the advisory itself did not make that clear.
Rohit Kumar, founder of The Quantum Hub, a policy think tank that has worked with large Artificial Intelligence startups, welcomed the reversal. The March 1 advisory “would have severely reduced speed to market and dented the innovation ecosystem,” Mr. Kumar said. “While the revision is definitely a positive step, the whole episode highlights the need for procedural safeguards — to avoid quick reactions to incidents and instead adopt a more consultative approach to policymaking.”
