India is signalling a major shift in how it governs artificial intelligence. A proposed new law could introduce a risk-based vetting system, creating a separate, stricter safety framework for the most powerful and advanced AI models before their public
release.
A New Chapter in AI Governance
For years, India has taken a relatively light-touch approach to AI, relying on existing laws and advisories to manage its growth. However, recent government statements suggest this is about to change. Union Electronics and IT Minister Ashwini Vaishnaw has confirmed that a consultation paper on AI regulation will be released in the coming weeks, noting that the risks associated with AI have 'crossed a threshold' warranting new, dedicated legislation. This marks a deliberate pivot from a technology-neutral stance to one of direct oversight, especially for what are known as 'frontier models'—the most advanced systems with the greatest potential for both progress and harm.
What is 'Risk-Based' Vetting?
The core idea behind the proposed law is not to treat all AI the same. Instead, it aims to create a tiered system based on risk. Low-risk applications, such as basic chatbots or productivity tools, would likely face minimal regulation. The real focus is on high-risk and frontier AI models. These are systems with vast capabilities, such as those that could discover software vulnerabilities, enable sophisticated cyberattacks, or generate highly convincing deepfakes. For these powerful models, the government is considering a mandatory vetting process. This could include pre-release assessments, controlled testing in a 'sandbox' environment to identify vulnerabilities, and security evaluations that are proportional to a model's capabilities. Developers might be required to prove their systems have adequate safeguards before they can be released to the public.
Defining the 'Frontier'
One of the most significant challenges for regulators will be defining what constitutes a 'frontier model' that requires this special scrutiny. The classification could be based on the sheer amount of computing power used to train a model, a common metric for identifying the largest and most complex systems. However, officials recognize that capability-based assessments may also be necessary, as a model's potential impact isn't always tied directly to its size. The proposed law would look beyond just the content an AI generates and instead scrutinize the underlying capabilities of the system itself. This distinction is crucial for addressing risks like autonomous AI agents acting without authorization or models being used to harm critical infrastructure.
Balancing Innovation and Safety
The government's proposed approach tries to strike a delicate balance. By focusing intense scrutiny only on the highest-risk models, the framework aims to avoid stifling innovation for the thousands of startups and developers working on less powerful, everyday AI applications. For the broader tech industry, this could increase trust in AI solutions and even become a competitive advantage. However, it will also likely lead to higher development and compliance costs for companies building frontier models. These firms will need to work more closely with legal and compliance teams and invest in robust systems for bias detection, ethical reviews, and transparent documentation. The ultimate goal is to create a predictable environment that fosters responsible innovation while protecting national security and public safety.
India's Place in the Global AI Conversation
India's move toward a risk-based framework mirrors similar efforts globally, most notably the EU's AI Act. However, its specific focus on pre-release vetting for frontier models also shows it is drawing lessons from discussions in the United States, where national security concerns have led to restrictions on certain advanced AI systems. By creating a formal, two-track system—one for general AI and another for the high-risk frontier—India is positioning itself to be a key player in the global conversation on AI governance. The forthcoming consultation paper will be a critical first step, seeking feedback from industry, civil society, and citizens to shape a law that is both effective and practical for India's booming digital economy.
