India is charting its own path in the global conversation on artificial intelligence, preparing a dedicated legal framework to govern the technology. This move signals a shift towards direct oversight of the most powerful AI systems.
A Shift in Strategy
India is moving from
a largely hands-off, technology-neutral stance to developing a specific legal framework for artificial intelligence. Recent statements from government officials, including Union Electronics and IT Minister Ashwini Vaishnaw, suggest a growing belief that the risks associated with AI have reached a point where new legislation is necessary. This marks a significant change, reflecting a global trend toward managing the societal and security implications of rapidly advancing AI. The government is expected to release a consultation paper to gather feedback from citizens, industry, and other stakeholders before drafting a new law. This planned framework is expected to exist alongside, and build upon, existing regulations like the IT Act and the Digital Personal Data Protection Act, 2023.
Focus on 'Frontier Models'
The core of the proposed legislation will likely focus on what are known as 'frontier models'. These are the most advanced, cutting-edge AI systems with capabilities that could pose risks to national security, cybersecurity, or societal stability if released without proper safeguards. Instead of applying the same heavy compliance burden on all AI developers, the government aims to create a distinction between general AI applications and these high-risk frontier systems. This risk-based approach would allow regulators to apply enhanced scrutiny only where it is most needed, a method favored by several large technology companies. The challenge, however, will be defining what qualifies as a frontier model, with officials considering factors like the computing power used for training as well as the model's specific capabilities.
What Conditions Could Apply?
For developers of these high-risk models, the new framework could introduce several pre-deployment conditions. These may include mandatory safety assessments, controlled testing in secure environments, and security evaluations to prove that appropriate protections are in place before a public launch. The framework is also expected to address threats that go beyond simple content moderation, such as the potential for AI agents to take unauthorized actions, identify vulnerabilities in critical infrastructure, or generate harmful instructions. Human oversight is emerging as another central pillar, with potential requirements for mandatory human approval for sensitive actions, continuous monitoring, and emergency 'kill switches' to stop unsafe AI behavior.
Balancing Innovation and Safety
The government's stated goal is to strike a balance between fostering India's burgeoning AI ecosystem and mitigating potential harms. India has so far opted for a 'pro-innovation' or 'light-touch' approach, using existing laws and advisories rather than a single, heavy-handed AI law like the European Union's AI Act. This new framework appears to continue that philosophy by targeting only the highest-risk models, thereby avoiding stifling smaller startups and general AI development. This approach aligns with the 'IndiaAI Mission', which seeks to build sovereign AI capabilities and promote AI as a tool for economic growth. The emphasis is on creating a 'techno-legal' solution where technology and law work together, with the industry shouldering significant responsibility for managing risks.
Addressing Deepfakes and Misinformation
The planned regulations are also a direct response to the rising tide of AI-generated misinformation, particularly deepfakes. Recent amendments to India's IT Rules have already introduced a stricter regime for what is termed 'Synthetically-Generated Information' (SGI). These rules mandate clear labeling of AI-generated content and have drastically shortened the timeline for platforms to take down unlawful content, including deepfakes, to just a few hours after receiving a government or court order. The forthcoming AI law is expected to build on these measures, ensuring clearer identification of synthetic media and establishing accountability for harm caused by such content.
