Safety regulations for artificial intelligence (AI) should be formed through a consultation process involving India, and not unilaterally by the two sectoral leaders — China and the US, Sudeep Shrivastava, chief operating officer of the IndiaAI Mission, said at India Mobile Congress 2026 here in New Delhi.
“Recently there was a meeting between the leadership of the two countries that are leading on this (AI) front, US and China. But it is very important that India, which is one of the leaders of Global South, and also other countries of the world, are also part of this consultation process through which the regulations come up,” Shrivastava said, highlighting the role of India in the process of building AI safety regulations.
The remarks follow a state visit in late September by Chinese President Xi Jinping to Washington, where he met his American counterpart, Donald Trump. Both parties agreed to establish an AI safety and incident communication channel while also agreeing to hold a dialogue on the technology’s risks and benefits, with the next round of discussions set for November.
Shrivastava added that 13 projects are already in place under the Safe and Trusted AI pillar, with the mission intending to increase the number following the formal constitution of the AI Safety Institute.
However, Mayank Vatsa, a senior professor at IIT Jodhpur whose research focuses on artificial intelligence, raised concerns about India’s lack of indigenous AI safety benchmarks.
“The world’s largest, most populous country has not been able to produce a single AI safety benchmark till date,” Vatsa said, adding that none of the scientific models use benchmarks that have been developed independently in India.
Meanwhile, Shrivastava said that unlike the Industrial Revolution, in the AI revolution “things are changing on a week-to-week basis”, which has made innovation one of the prime focuses of the IndiaAI Mission.
He also highlighted the release of Claude’s Mythos, a highly restricted frontier model developed by Anthropic, and the attacks by OpenAI AI agents on the computer systems of another AI platform, Hugging Face. Talking about these incidents, he said there is a dire need to have proper regulation and guardrails in place for AI safety.