Reining in AI

The IT ministry’s recent AI advisory has only made the debate over regulations more acute

 amid the ongoing debates surrounding the regulation of artificial intelligence (AI), a recent advisory from the Ministry of Electronics and Information Technology (MeitY) marks a potential turning point. Issued on March 1, it requires AI companies and platforms to obtain explicit permission from the government before launching AI models, particularly those still in the testing phase. This directive follows a notification in December last year, which expressed concerns regarding the increasing prevalence of deepfakes on social media platforms.

The notification prompted AI firms, especially start-ups, to express their discontent on social media. The advisory was also criticised by investors like Martin Casado of venture capital fund Andreessen Horowitz, who called it “anti-innovation”.

However, Rajeev Chandrasekhar, Minister of State for Skill Development & Entrepreneurship and Electronics & Information Technology clarified on X that the advisory targeted only big tech companies. On March 15, the government issued a new advisory stipulating that AI models should only be accessible to Indian users provided they have clear labels indicating potential inaccuracies or unreliability in the output they generate. But, industry watchers say the call for monitoring AI has only intensified. Experts also emphasise the ambiguity of AI regulation, underscoring the challenge of navigating this uncharted territory. “This debate over how to tackle AI, especially generative AI (Gen AI), is taking place all over the world and not just in India. We are still grappling with this new beast,” says Neil Shah, Partner and Co-founder of Counterpoint Research, a market research firm.

Importantly, Chandrasekhar made a distinction between big tech firms that are using Gen AI and start-ups. Shah points out that this distinction was necessary because the data that is needed to train Gen AI platforms is usually stored either in the US or in China. “So, this data is not India-specific; which is why you have a scenario where if you ask an AI about the map of India, it gives you an inaccurate picture. That is a problem,” he says.

This scenario is not prevalent with Indian start-ups due to their limited data usage and smaller scale of deployment. “The Gen AI used by big tech companies can be integrated into other platforms, like smartphones. So, the use case is huge. Indian start-ups have not achieved that scale,” he explains.

For the time being, it seems that Indian start-ups can heave a sigh of relief. But, what shape a future regulatory framework for AI will take, is something only time will tell.