New Delhi:
Ajay Kumar Sood, the Principal Scientific Adviser to the Centre, has said the government is open to the idea of introducing a dedicated law on artificial intelligence, a statement that comes amid a growing unease over AI harms.
In an exclusive interview with NDTV, Sood also said the Centre has already initiated the process of preparing a consultation paper on AI regulations.
He emphasised that while AI presents unprecedented opportunities, growing risks demand carefully designed safeguards and governance frameworks.
"What has happened in the last one or two years, things have moved really much ahead than what they were before. And the risks are becoming more and more clear that we need to do something. So, the idea is to have targeted regulatory guardrails," he said.
Researchers and safety experts at leading AI companies, including OpenAI and Anthropic, have recently flagged stark warnings about the potential risks from increasingly powerful frontier models.
READ | OpenAI Says Its AI Agents Weakened Parts Of An Internal Monitoring System
India has been tightening its IT rules to crack down on AI content. In February, the Centre brought in stricter obligations for online platforms on handling AI-generated and synthetic content, including deepfakes, saying platforms, such as X and Instagram, must take down within three hours any such content flagged by a competent authority or court. The government notified amendments to the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021, that formally define AI-generated and synthetic content.
"We are working on this theme right from the beginning that how AI has to be useful to our society but at the same time we have to avoid the obvious risks," Sood told NDTV.
READ | AI Companies Prepare For Public Backlash After Potential Catastrophic Incident: Report
He also said the agentic AIs have posed a "new challenge", reffering to two OpenAI models that escaped from a closed testing environment and autonomously broke into the internal systems of Hugging Face, a site AI developers use to store and share code.
"These agentic AIs, if left on their own, can do more harm," Sood said.
READ | "Act With Urgency": Microsoft Chief Calls For Effective AI Safety Standards
He said the Indian Computer Emergency Response Team (CERT-In), the government's cybersecurity arm, has also issued an advisory that to deploy agentic AI safely, a human must be kept in the loop to approve critical actions.
The CERT-In has also that every agent should be isolated in sandboxed environments, he said.
How may i help you today