A Shift from a Hands-Off Approach
For years, India's official stance on AI has been to foster growth with minimal interference. The government preferred to let the industry innovate, relying on existing laws and voluntary guidelines, such as the India AI Governance Guidelines of November
2025, rather than introducing a specific AI law. This was often positioned as a competitive advantage against more restrictive regimes like the European Union's. However, that is set to change. Union Minister for Electronics and IT, Ashwini Vaishnaw, has announced that a consultation paper will be released in the coming weeks, likely in November 2026, to form the basis of a new regulatory framework. This move acknowledges that the risks associated with AI, from deepfakes to algorithmic bias, have crossed a threshold where formal legal guardrails are becoming necessary.
What 'Model Testing and Release' Means
At the heart of the new discussions are the concepts of 'model testing' and 'release'. Think of an AI model as an engine. In a lab, it can be powerful, but its real-world impact is unknown. Pre-release testing involves rigorously evaluating this engine before it's put into a car and given to the public. This means checking for flaws, biases, and potential for misuse in a controlled environment. The government is considering several layers of scrutiny. These include pre-release evaluations to assess a model’s capabilities, testing in secure 'sandboxes' to see how it behaves under supervision, and assigning clear responsibility for any harm it might cause after its public release. The aim is to catch dangerous capabilities, such as the ability to generate sophisticated cyberattacks or highly convincing synthetic content, before they can be widely exploited.
Balancing Innovation with Safety
The government's challenge is to craft rules that ensure safety without stifling the growth of India's burgeoning AI startup ecosystem. Officials are exploring a risk-based framework, which means not all AI systems would be treated equally. The strictest rules would likely apply to 'frontier models'—the most powerful and capable AI systems that pose the greatest potential risks. This could create a two-track system: one with tough pre-release checks for high-risk AI, and another with more relaxed rules for lower-risk applications. This approach seeks a middle path, balancing domestic safety needs with international practices, ensuring that regulation doesn't overburden legitimate research while still addressing serious risks. The consultation will solicit feedback from industry, civil society, and the public to get this balance right.
The Global Context for India's AI Law
India's move toward formal AI legislation is part of a global trend. Governments worldwide are grappling with how to govern this powerful technology. The European Union has its comprehensive, risk_based AI Act, which is already entering its enforcement phase. China has implemented targeted rules focused on social stability and content control. The United States has a more fragmented approach, with a mix of federal executive orders and a growing number of state_level laws. India's planned consultation suggests it is trying to forge its own path, drawing lessons from these different approaches. The goal appears to be a framework that is uniquely suited to India’s economic ambitions and societal needs, establishing human_centric principles and ensuring fail_safe human control over AI systems.
