The Ministry of Electronics and Information Technology (MeitY) has been deliberating over the institute’s objectives, budget, and framework for some time
The Indian government is taking steps towards establishing a national AI Safety Institute (AISI) to address emerging challenges in artificial intelligence. Speaking at Microsoft’s “Building AI Companions for India” event in Bangalore, MeitY Secretary S Krishnan discussed the need for proactive AI regulation alongside Microsoft AI CEO Mustafa Suleyman.
“AI Safety Institute seems to be the flavour of the day across the world, and we are in the process of trying to establish one ourselves to understand this better,” Krishnan said. He emphasised the importance of balanced regulation, noting, “We’ve often waited for things to go wrong. It’s time to be thoughtful and deliberate, rather than treating regulation as a taboo.”
The Ministry of Electronics and Information Technology (MeitY) has been deliberating over the institute’s objectives, budget, and framework for some time. A meeting held last month focused on charting the institute’s key goals. Krishnan pointed out that while existing legislation has been effective in addressing concerns like misrepresentation and deepfakes during elections, the evolving nature of AI demands ongoing vigilance.
“With issues like misrepresentation and deepfakes, we feared significant disruptions, especially during the Indian elections and globally in 2024. However, existing legislation has proven reasonably effective in mitigating those risks,” he said.
However, Krishnan acknowledged that AI remains in its early stages, with much still to unfold. Mustafa Suleyman highlighted the unpredictable nature of AI’s evolution. “The tricky part for us in the next five years is figuring out how models with the ability to improve themselves independently—recursive self-improvement—will develop,” Suleyman said, emphasising the need for adaptive regulatory approaches.
Global Push For AI Safety Standards
Globally, there is growing momentum to create AI safety frameworks. The UK was the first to launch an AI Safety Institute, focusing on research and testing advanced AI models. Companies like OpenAI, Meta, Google DeepMind, and Microsoft have signed voluntary agreements granting the UK early access to their AI models. Ian Hogarth, chair of the UK Government’s AI Foundation Model Taskforce, highlighted the significance of these efforts. “We hosted the Bletchley AI Safety Summit and secured agreements between China, the US, and Europe to address these risks,” Hogarth said, pointing to the importance of international cooperation.
In the US, companies such as OpenAI and Anthropic signed MOUs with the US AI Safety Institute earlier this year. Elizabeth Kelly, director of the institute, stressed the value of safety in fostering innovation. “Safety promotes trust, which promotes adoption, driving innovation,” she explained.
China has not yet formally committed to an AI Safety Institute, but reports indicate that influential voices are advocating for its creation. Meanwhile, Singapore has designated the Digital Trust Centre at Nanyang Technological University as its AI Safety Institute.
The EU introduced the AI Act earlier this year, the world’s first comprehensive legal framework for AI. While aimed at ensuring safety and trust, some critics argue it may stifle innovation by creating regulatory barriers for AI developers.
As India moves closer to establishing its AI Safety Institute, it joins a global effort to ensure AI development remains safe, transparent, and beneficial. The initiative reflects a growing recognition that thoughtful regulation is essential to managing the risks associated with rapidly advancing AI technologies.
